mirror/hdf5
2
0
Fork 0
mirror of https://github.com/HDFGroup/hdf5.git synced 2024-12-09 07:32:32 +08:00
Code Issues Packages Projects Releases Wiki Activity
1b1e1ebbbd
hdf5/src/H5G.c

3767 lines
125 KiB
C
Raw Normal View History

[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
/* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* Copyright by the Board of Trustees of the University of Illinois. *
* All rights reserved. *
* *
* This file is part of HDF5. The full HDF5 copyright notice, including *
* terms governing use, modification, and redistribution, is contained in *
* the files COPYING and Copyright.html. COPYING can be found at the root *
* of the source code distribution tree; Copyright.html can be found at the *
* root level of an installed copy of the electronic HDF5 document set and *
* is linked from the top-level documents page. It can also be found at *
* http://hdf.ncsa.uiuc.edu/HDF5/doc/Copyright.html. If you do not have *
* access to either file, you may request a copy from hdfhelp@ncsa.uiuc.edu. *
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
[svn-r2] Oops...
1997-07-31 05:17:56 +08:00
/*-------------------------------------------------------------------------
*
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
* Created: H5G.c
* Jul 18 1997
* Robb Matzke <matzke@llnl.gov>
*
* Purpose: Symbol table functions. The functions that begin with
* `H5G_stab_' don't understand the naming system; they operate
* on a single symbol table at a time.
*
* The functions that begin with `H5G_node_' operate on the leaf
* nodes of a symbol table B-tree. They should be defined in
* the H5Gnode.c file.
*
* The remaining functions know how to traverse the group
* directed graph.
*
* Names: Object names are a slash-separated list of components. If
* the name begins with a slash then it's absolute, otherwise
* it's relative ("/foo/bar" is absolute while "foo/bar" is
* relative). Multiple consecutive slashes are treated as
* single slashes and trailing slashes are ignored. The special
* case `/' is the root group. Every file has a root group.
*
* API functions that look up names take a location ID and a
* name. The location ID can be a file ID or a group ID and the
* name can be relative or absolute.
*
* +--------------+----------- +--------------------------------+
* | Location ID | Name | Meaning |
* +--------------+------------+--------------------------------+
* | File ID | "/foo/bar" | Find `foo' within `bar' within |
* | | | the root group of the specified|
* | | | file. |
* +--------------+------------+--------------------------------+
* | File ID | "foo/bar" | Find `foo' within `bar' within |
* | | | the current working group of |
* | | | the specified file. |
* +--------------+------------+--------------------------------+
* | File ID | "/" | The root group of the specified|
* | | | file. |
* +--------------+------------+--------------------------------+
* | File ID | "." | The current working group of |
* | | | the specified file. |
* +--------------+------------+--------------------------------+
* | Group ID | "/foo/bar" | Find `foo' within `bar' within |
* | | | the root group of the file |
* | | | containing the specified group.|
* +--------------+------------+--------------------------------+
* | Group ID | "foo/bar" | File `foo' within `bar' within |
* | | | the specified group. |
* +--------------+------------+--------------------------------+
* | Group ID | "/" | The root group of the file |
* | | | containing the specified group.|
* +--------------+------------+--------------------------------+
* | Group ID | "." | The specified group. |
* +--------------+------------+--------------------------------+
[svn-r2] Oops...
1997-07-31 05:17:56 +08:00
*
*
[svn-r15] ./src/H5AC.c ./src/H5ACprivate.h ./src/H5ACproto.h ./src/H5Bproto.h ./src/H5Gproto.h ./src/H5Hproto.h ./src/H5MFprivate.h ./src/H5MFproto.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5MMproto.h Changed my e-mail address. ./src/H5O.c NEW ./src/H5Onull.c NEW ./src/H5Ocont. NEW ./src/H5Ostab.c NEW New functions for dealing with object headers. The H5O.c is the generic stuff, and each particular message has a source file for the specific stuff. Use ./src/H5Ostab.c as a model for implementing other messages. ./src/Makefile ./src/test/Makefile Added new files ./src/debug.c Added debugging calls for object headers. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c ./src/H5MF.c Changed my e-mail address. Improved error handling. ./src/H5Eprivate.h ./src/H5Eproto.h Added more error symbols ./src/H5F.c Changed my e-mail address. Used macros for sizeof offsets and lengths. Added the interface initialization function to H5F_block_read() and H5F_block_write(). Updated H5F_debug() ./src/H5Fprivate.h Got rid of H5F_symbol_table_size(). Use H5G_SIZEOF_ENTRY() instead. Reformatted H5F_decode_offset() for readability. ./src/H5G.c Changed my e-mail address. Improved error handling. Replaced not_implemented_yet__*() with real functions from H5O. ./src/H5Gprivate.h Changed `symtab' to `stab' to be consistent with other stuff. ./src/H5H.c ./src/H5Hprivate.h Changed my e-mail address. Improved error handling. Added an extra argument to H5H_new() to indicate whether you want a global heap or a local heap. ./src/hdf5gen.h Added NELMTS() Fixed FUNC_ENTER() Rewrote FUNC_LEAVE() in terms of HRETURN(). ./src/hdf5meta.h Added `const' to the decode macros. ./src/hdf5pabl.h Added PABLO_SAVE() ./test/testhdf5.c ./test/testhdf5.h Added calls for object header testing. ./test/theap.c Turned off some output.
1997-08-08 03:23:00 +08:00
* Modifications:
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Robb Matzke, 5 Aug 1997
* Added calls to H5E.
[svn-r2] Oops...
1997-07-31 05:17:56 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Robb Matzke, 30 Aug 1997
* Added `Errors:' field to function prologues.
[svn-r61] ./src/H5.c Changed atexit() to HDatexit(). ./src/H5AC.c Changed qsort() to HDqsort(). ./src/H5B.c Changed memcmp() to HDmemcmp(). ./src/H5E.c ./src/H5M.c Added parentheses around indirect function calls so that checkposix doesn't complain and to advertise that it's an indirect function call. ./src/H5F.c ./src/H5G.c Added errors field to each function prologue to test the script mentioned above. ./src/H5H.c Changed fputc() to HDfputc(). ./src/H5MM.c ./src/H5MMprivate.h The argument for H5MM_xfree() is a pointer to a constant even though the argument to free() isn't. This reduces the number of warnings in other parts of the code but creates a new warning in this file. ./src/H5O.c Plugged a few memory leaks that happen during error handling. ./src/H5Oname.c ./src/debug.c Added `HD' to the beginning of some posix functions.
1997-09-02 23:38:26 +08:00
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Pedro Vicente, 22 Aug 2002
* Added `id to name' support.
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
*
[svn-r2] Oops...
1997-07-31 05:17:56 +08:00
*-------------------------------------------------------------------------
*/
[svn-r2652] Purpose: Maintainance & performance enhancements Description: Re-arranged header files to protect private symbols better. Changed optimized regular hyperslab I/O to compute the offsets more efficiently from previous method of using matrix operations. Added sequential I/O operations at a more abstract level (at the same level as H5F_arr_read/write), to support the optimized hyperslab I/O. Platforms tested: Solaris 2.6 (baldric) & FreeBSD 4.1.1 (hawkwind)
2000-10-10 15:43:38 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
#define H5G_PACKAGE /*suppress error message about including H5Gpkg.h */
#define H5F_PACKAGE /*suppress error about including H5Fpkg */
[svn-r2652] Purpose: Maintainance & performance enhancements Description: Re-arranged header files to protect private symbols better. Changed optimized regular hyperslab I/O to compute the offsets more efficiently from previous method of using matrix operations. Added sequential I/O operations at a more abstract level (at the same level as H5F_arr_read/write), to support the optimized hyperslab I/O. Platforms tested: Solaris 2.6 (baldric) & FreeBSD 4.1.1 (hawkwind)
2000-10-10 15:43:38 +08:00
[svn-r2] Oops...
1997-07-31 05:17:56 +08:00
/* Packages needed by this file... */
[svn-r6837] Purpose: Code cleanup. Description: Move many package or internal function prototypes and macro definitions into tighter scope according to their current use. Added more comments where appropriate. Eliminate ancient, unused functions. Added a couple "accessor" functions to get parts of data structures which were moved out of scope. Platforms tested: h5committested
2003-05-09 04:09:07 +08:00
#include "H5private.h" /* Generic Functions */
#include "H5Aprivate.h" /* Attributes */
[svn-r6878] Purpose: Code cleanup Description: Limit the scope on more function prototypes/macros/typedefs. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest not necessary.
2003-05-16 03:22:33 +08:00
#include "H5Bprivate.h" /* B-link trees */
#include "H5Dprivate.h" /* Datasets */
[svn-r6837] Purpose: Code cleanup. Description: Move many package or internal function prototypes and macro definitions into tighter scope according to their current use. Added more comments where appropriate. Eliminate ancient, unused functions. Added a couple "accessor" functions to get parts of data structures which were moved out of scope. Platforms tested: h5committested
2003-05-09 04:09:07 +08:00
#include "H5Eprivate.h" /* Error handling */
#include "H5Fpkg.h" /* File access */
#include "H5FLprivate.h" /* Free Lists */
#include "H5Gpkg.h" /* Groups */
#include "H5HLprivate.h" /* Local Heaps */
#include "H5Iprivate.h" /* IDs */
#include "H5MMprivate.h" /* Memory management */
#include "H5Oprivate.h" /* Object headers */
[svn-r15] ./src/H5AC.c ./src/H5ACprivate.h ./src/H5ACproto.h ./src/H5Bproto.h ./src/H5Gproto.h ./src/H5Hproto.h ./src/H5MFprivate.h ./src/H5MFproto.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5MMproto.h Changed my e-mail address. ./src/H5O.c NEW ./src/H5Onull.c NEW ./src/H5Ocont. NEW ./src/H5Ostab.c NEW New functions for dealing with object headers. The H5O.c is the generic stuff, and each particular message has a source file for the specific stuff. Use ./src/H5Ostab.c as a model for implementing other messages. ./src/Makefile ./src/test/Makefile Added new files ./src/debug.c Added debugging calls for object headers. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c ./src/H5MF.c Changed my e-mail address. Improved error handling. ./src/H5Eprivate.h ./src/H5Eproto.h Added more error symbols ./src/H5F.c Changed my e-mail address. Used macros for sizeof offsets and lengths. Added the interface initialization function to H5F_block_read() and H5F_block_write(). Updated H5F_debug() ./src/H5Fprivate.h Got rid of H5F_symbol_table_size(). Use H5G_SIZEOF_ENTRY() instead. Reformatted H5F_decode_offset() for readability. ./src/H5G.c Changed my e-mail address. Improved error handling. Replaced not_implemented_yet__*() with real functions from H5O. ./src/H5Gprivate.h Changed `symtab' to `stab' to be consistent with other stuff. ./src/H5H.c ./src/H5Hprivate.h Changed my e-mail address. Improved error handling. Added an extra argument to H5H_new() to indicate whether you want a global heap or a local heap. ./src/hdf5gen.h Added NELMTS() Fixed FUNC_ENTER() Rewrote FUNC_LEAVE() in terms of HRETURN(). ./src/hdf5meta.h Added `const' to the decode macros. ./src/hdf5pabl.h Added PABLO_SAVE() ./test/testhdf5.c ./test/testhdf5.h Added calls for object header testing. ./test/theap.c Turned off some output.
1997-08-08 03:23:00 +08:00
[svn-r6878] Purpose: Code cleanup Description: Limit the scope on more function prototypes/macros/typedefs. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest not necessary.
2003-05-16 03:22:33 +08:00
/* Local macros */
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
#define H5G_INIT_HEAP 8192
#define H5G_RESERVED_ATOMS 0
[svn-r6878] Purpose: Code cleanup Description: Limit the scope on more function prototypes/macros/typedefs. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest not necessary.
2003-05-16 03:22:33 +08:00
#define H5G_SIZE_HINT 256 /*default root grp size hint */
#define H5G_NLINKS 16 /*max symlinks to follow per lookup */
/*
* During name lookups (see H5G_namei()) we sometimes want information about
* a symbolic link or a mount point. The normal operation is to follow the
* symbolic link or mount point and return information about its target.
*/
#define H5G_TARGET_NORMAL 0x0000
#define H5G_TARGET_SLINK 0x0001
#define H5G_TARGET_MOUNT 0x0002
[svn-r6837] Purpose: Code cleanup. Description: Move many package or internal function prototypes and macro definitions into tighter scope according to their current use. Added more comments where appropriate. Eliminate ancient, unused functions. Added a couple "accessor" functions to get parts of data structures which were moved out of scope. Platforms tested: h5committested
2003-05-09 04:09:07 +08:00
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
#define PABLO_MASK H5G_mask
[svn-r15] ./src/H5AC.c ./src/H5ACprivate.h ./src/H5ACproto.h ./src/H5Bproto.h ./src/H5Gproto.h ./src/H5Hproto.h ./src/H5MFprivate.h ./src/H5MFproto.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5MMproto.h Changed my e-mail address. ./src/H5O.c NEW ./src/H5Onull.c NEW ./src/H5Ocont. NEW ./src/H5Ostab.c NEW New functions for dealing with object headers. The H5O.c is the generic stuff, and each particular message has a source file for the specific stuff. Use ./src/H5Ostab.c as a model for implementing other messages. ./src/Makefile ./src/test/Makefile Added new files ./src/debug.c Added debugging calls for object headers. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c ./src/H5MF.c Changed my e-mail address. Improved error handling. ./src/H5Eprivate.h ./src/H5Eproto.h Added more error symbols ./src/H5F.c Changed my e-mail address. Used macros for sizeof offsets and lengths. Added the interface initialization function to H5F_block_read() and H5F_block_write(). Updated H5F_debug() ./src/H5Fprivate.h Got rid of H5F_symbol_table_size(). Use H5G_SIZEOF_ENTRY() instead. Reformatted H5F_decode_offset() for readability. ./src/H5G.c Changed my e-mail address. Improved error handling. Replaced not_implemented_yet__*() with real functions from H5O. ./src/H5Gprivate.h Changed `symtab' to `stab' to be consistent with other stuff. ./src/H5H.c ./src/H5Hprivate.h Changed my e-mail address. Improved error handling. Added an extra argument to H5H_new() to indicate whether you want a global heap or a local heap. ./src/hdf5gen.h Added NELMTS() Fixed FUNC_ENTER() Rewrote FUNC_LEAVE() in terms of HRETURN(). ./src/hdf5meta.h Added `const' to the decode macros. ./src/hdf5pabl.h Added PABLO_SAVE() ./test/testhdf5.c ./test/testhdf5.h Added calls for object header testing. ./test/theap.c Turned off some output.
1997-08-08 03:23:00 +08:00
[svn-r139] ./src/*.[ch] Removed the interface initialization argument from FUNC_ENTER() and made it a locally-defined preprocessor symbol, INTERFACE_INIT. Changed `offset' to `address' and `length' to `size' in documentation so it's more consistent. `Offset' still appears occassionally when it refers to a byte offset within some other data structure. Moved interface termination function prototypes from public header files to .c files and made them static. ./src/H5.c ./src/H5public.h Added H5init() because it's possible that the predefined data types are not initialized. This happens only if the first call to the hdf5 library passes a predefined data type symbol as an argument. There should be some way to fix this... ./src/H5A.c ./src/H5Aprivate.h ./src/H5Apublic.h The free_func returns SUCCEED or FAIL, although the return value is ignored by H5A. This is so we can use the various H5*_close() functions to free things. H5Ainc_ref() and H5Adec_ref() are no longer public. Many of the other atom functions should also be made private, but I'll save that for later... Added additional template groups called H5_TEMPLATE_0 through H5_TEMPLATE_7 that are used by the various template subclasses. Increased the number of bits used for atom groups to prevent negative atoms. ./src/H5AC.c ./src/H5ACprivate.h Changed H5AC_new() to H5AC_create() to make names more consistent. ./src/H5B.c ./src/H5Bprivate.h Changed H5B_new() to H5B_create() to make names more consistent. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Now supports multiple subclasses of templates, although it's done with big switch statements. The default values for templates are defined in the source file to which that template belongs. This got rid of lots of needless preprocessor constants. Added H5Ccreate() to create a new template. Changed H5C_release() to H5Cclose() to make the naming more consistent. ./src/H5D.c ./src/H5Dprivate.h ./src/H5Dpublic.h Enhanced to use the new dataset interface, and uses the enhanced data type and data space interfaces, which haven't been completely implemented. The dataset interface doesn't handle non-contiguous storage, compression, or data type and space conversions yet. ./src/H5F.c ./src/H5Fprivate.h ./src/H5Fpublic.h Removed H5Fflush() since just calls H5F_flush(), which doesn't do what the user would probably think it does, namely, flush everything. It only flushes those things sitting in the H5AC cache and the boot block. Changed the `file_create_parms' field of H5F_low_t to just `create_parms' since the `file' part is obvious. ./src/H5Fistore.c Added some support for external files. Mostly just in the file format and not supported much by the library yet. I need to finish some dataset functions first. Changed H5F_istore_new() to H5F_istore_create() to make names more uniform across packages. ./src/H5Flow.c Flushing a file causes the file to be physically extended to the logical eof. This prevents H5F_open() from thinking a file has been truncated. Most of the time the file will already be that large, and when it isn't Unix will often just allocate the final block anyway. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Removed H5G_basename() Removed (temporarily) data type information from symbol table entries and renamed H5G_CACHED_SDATA to H5G_CACHED_SDSPACE to reflect that it's a simple data space and has nothing to do with raw data. Changed H5G_node_new() to H5G_node_create() and H5G_stab_new() to H5G_stab_create() to make names more uniform across packages. Fixed an undefined address bug that happens when H5G_node_debug() program doesn't pass enough info to H5G_node_load(). ./src/H5H.c ./src/H5Hprivate.h Changed H5H_new() to H5H_create() to make the names more uniform across packages. ./src/H5M.c ./src/H5Mprivate.h ./src/H5Mpublic.h Nulled all the create functions. Most of the other callbacks are to public functions. Removed H5Mcreate(). Changed hobjtype_t to group_t since it has to be the same thing anyway. ./src/H5O.c ./src/H5Oprivate.h ./src/H5Osdim.c ./src/H5Osdtyp.c Changed H5O_SIM_DIM to H5O_SDSPACE (simple data space) since `simple data space' is its official name, not `simple dimensions'. Will eventually add H5O_CDSPACE for comples data spaces. Changed _sim_dim_ to _dspace_. Replaced H5O_SIM_DTYPE and the compound data type messages with a single H5O_DTYPE message. Changed _sim_dtype_ to _dtype_. Changed H5O_STD_STORE to H5O_CSTORE (contiguous storage) since contiguous storage is not necessarily standard. Changed _std_store_ to _cstore_ in H5Ocstore.c Added the H5O_EFL (external file list) message. Changed H5O_new() to H5O_create() to make names more uniform across packages. ./src/H5Oefl.c NEW External file list message for specifying which non-hdf5 files contain raw data for a dataset. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Renamed and moved data structures to make the names conform to our naming scheme. ./src/H5T.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5Tpkg.h NEW Data structures redesigned to be more flexible. The interface was redesigned to make it more regular and to make some names more uniform across packages. ./src/H5detect.c Output was changed to produce a file that conforms to the hdf5 coding standard. ./src/Makefile.in Generates H5Tinit.c by running H5detect. ./src/debug.c Moved command argument processing.
1997-12-11 06:41:07 +08:00
/* Interface initialization */
[svn-r4355] Purpose: Code cleanup (sorta) Description: When the first versions of the HDF5 library were designed, I remembered vividly the difficulties of porting code from a 32-bit platform to a 16-bit platform and asked that people use intn & uintn instead of int & unsigned int, respectively. However, in hindsight, this was overkill and unnecessary since we weren't going to be porting the HDF5 library to 16-bit architectures. Currently, the extra uintn & intn typedefs are causing problems for users who'd like to include both the HDF5 and HDF4 header files in one source module (like Kent's h4toh5 library). Solution: Changed the uintn & intn's to unsigned and int's respectively. Platforms tested: FreeBSD 4.4 (hawkwind)
2001-08-15 06:09:56 +08:00
static int interface_initialize_g = 0;
[svn-r328] Changes since 19980318 ---------------------- ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gpublic.h ./src/H5Gstab.c Implemented H5Giterate(). However, since most functions can't take a group ID in place of a file ID yet, there's not a whole lot that the operator can do besides print the name or something. ./test/H5O.c Fixed writing of four uninitialized bytes to the file as part of an object header. ./test/istore.c For some reason, `mpirun -np 1 istore' results in extra arguments on the command line that istore doesn't understand. I'm probably forgetting to call some MPI function in main(). Albert, Kim? So I commented out the `exit(1)' for the time being.
1998-03-24 22:53:57 +08:00
#define INTERFACE_INIT H5G_init_interface
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
static herr_t H5G_init_interface(void);
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
[svn-r6878] Purpose: Code cleanup Description: Limit the scope on more function prototypes/macros/typedefs. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest not necessary.
2003-05-16 03:22:33 +08:00
/* Local typedefs */
[svn-r106] Fixed some symbol table bugs. Fixed a free memory read in H5Osdim.c.
1997-09-22 10:08:54 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Struct only used by change name callback function */
typedef struct H5G_names_t {
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
H5G_entry_t *loc;
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
H5RS_str_t *src_name;
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
H5G_entry_t *src_loc;
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
H5RS_str_t *dst_name;
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
H5G_entry_t *dst_loc;
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
H5G_names_op_t op;
} H5G_names_t;
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/* Enum for H5G_namei actions */
typedef enum {
H5G_NAMEI_TRAVERSE, /* Just traverse groups */
H5G_NAMEI_INSERT /* Insert entry in group */
[svn-r6878] Purpose: Code cleanup Description: Limit the scope on more function prototypes/macros/typedefs. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest not necessary.
2003-05-16 03:22:33 +08:00
} H5G_namei_act_t ;
/*
* This table contains a list of object types, descriptions, and the
* functions that determine if some object is a particular type. The table
* is allocated dynamically.
*/
typedef struct H5G_typeinfo_t {
[svn-r7192] Purpose: Code cleanup Description: Clean up various minor warnings in the library. Platforms tested: FreeBSD 4.8 (sleipnir) too small to need h5committest
2003-07-10 03:16:17 +08:00
H5G_obj_t type; /*one of the public H5G_* types */
[svn-r6878] Purpose: Code cleanup Description: Limit the scope on more function prototypes/macros/typedefs. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest not necessary.
2003-05-16 03:22:33 +08:00
htri_t (*isa)(H5G_entry_t*, hid_t); /*function to determine type */
char *desc; /*description of object type */
} H5G_typeinfo_t;
/* Local variables */
static H5G_typeinfo_t *H5G_type_g = NULL; /*object typing info */
static size_t H5G_ntypes_g = 0; /*entries in type table */
static size_t H5G_atypes_g = 0; /*entries allocated */
static char *H5G_comp_g = NULL; /*component buffer */
static size_t H5G_comp_alloc_g = 0; /*sizeof component buffer */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
[svn-r2476] Added free lists to track various data structures in memory, to reduce malloc abuse.
2000-08-17 04:13:02 +08:00
/* Declare a free list to manage the H5G_t struct */
H5FL_DEFINE(H5G_t);
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/* Declare extern the PQ free list for the wrapped strings */
H5FL_BLK_EXTERN(str_buf);
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
/* Private prototypes */
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
static H5G_t *H5G_create(H5G_entry_t *loc, const char *name, size_t size_hint, hid_t dxpl_id);
#ifdef NOT_YET
static H5G_t *H5G_reopen(H5G_t *grp);
#endif /* NOT_YET */
static htri_t H5G_isa(H5G_entry_t *ent, hid_t dxpl_id);
static herr_t H5G_link(H5G_entry_t *cur_loc, const char *cur_name,
H5G_entry_t *new_loc, const char *new_name,
H5G_link_t type, unsigned namei_flags, hid_t dxpl_id);
static herr_t H5G_linkval(H5G_entry_t *loc, const char *name, size_t size,
char *buf/*out*/, hid_t dxpl_id);
static herr_t H5G_move(H5G_entry_t *src_loc, const char *src_name,
H5G_entry_t *dst_loc, const char *dst_name, hid_t dxpl_it);
static herr_t H5G_unlink(H5G_entry_t *loc, const char *name, hid_t dxpl_id);
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
static herr_t H5G_get_num_objs(H5G_entry_t *grp, hsize_t *num_objs, hid_t dxpl_id);
static ssize_t H5G_get_objname_by_idx(H5G_entry_t *loc, hsize_t idx, char* name, size_t size, hid_t dxpl_id);
static H5G_obj_t H5G_get_objtype_by_idx(H5G_entry_t *loc, hsize_t idx, hid_t dxpl_id);
[svn-r7071] Purpose: Code cleanup Description: Tweak internal iterator over IDs to not require constant "operator" data. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-06-20 23:19:18 +08:00
static int H5G_replace_ent(void *obj_ptr, hid_t obj_id, void *key);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
static herr_t H5G_traverse_slink(H5G_entry_t *grp_ent/*in,out*/,
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
H5G_entry_t *obj_ent/*in,out*/, int *nlinks/*in,out*/, hid_t dxpl_id);
static herr_t H5G_set_comment(H5G_entry_t *loc, const char *name,
const char *buf, hid_t dxpl_id);
static int H5G_get_comment(H5G_entry_t *loc, const char *name,
size_t bufsize, char *buf, hid_t dxpl_id);
[svn-r7192] Purpose: Code cleanup Description: Clean up various minor warnings in the library. Platforms tested: FreeBSD 4.8 (sleipnir) too small to need h5committest
2003-07-10 03:16:17 +08:00
static herr_t H5G_register_type(H5G_obj_t type, htri_t(*isa)(H5G_entry_t*, hid_t),
[svn-r6878] Purpose: Code cleanup Description: Limit the scope on more function prototypes/macros/typedefs. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest not necessary.
2003-05-16 03:22:33 +08:00
const char *desc);
static H5G_t *H5G_rootof(H5F_t *f);
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
/*-------------------------------------------------------------------------
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Function: H5Gcreate
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
*
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
* Purpose: Creates a new group relative to LOC_ID and gives it the
* specified NAME. The group is opened for write access
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* and it's object ID is returned.
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* The optional SIZE_HINT specifies how much file space to
* reserve to store the names that will appear in this
* group. If a non-positive value is supplied for the SIZE_HINT
* then a default size is chosen.
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* See also: H5Gset(), H5Gpush(), H5Gpop()
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
*
* Errors:
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Return: Success: The object ID of a new, empty group open for
* writing. Call H5Gclose() when finished with
* the group.
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Failure: FAIL
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Programmer: Robb Matzke
* Wednesday, September 24, 1997
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
hid_t
[svn-r620] Changes since 19980825 ---------------------- ./MANIFEST ./src/H5R.c [NEW] ./src/H5Rprivate.h [NEW] ./src/H5Rpublic.h [NEW] ./src/Makefile.in ./src/hdf5.h ./test/ragged.c [NEW] Preliminary support for 2d ragged arrays for Mark Miller and Jim Reus. Not fully implemented yet. The test is not actually part of `make test' because we still have some memory problems. ./src/H5E.c ./src/H5Epublic.h Added H5E_RAGGED as a major error number. ./bin/release Checks the MANIFEST file against `svf ls' on systems that have it. ./bin/trace Fixed a bug that caused arguments of type `void *x[]' to not be handled. ./src/H5.c Removed unused variables and changed a couple types to fix compiler warnings. Added tracing support for ragged array object ID's and arrays of pointers. ./src/H5D.c H5Dcreate() will complain if either of the property lists are invalid (instead of using the default). ./src/H5D.c ./src/H5Dprivate.h Split H5Dget_space() into an API and internal function so it can be called from the new ragged array layer. ./src/H5Fistore.c Fixed warnings about unsigned vs. signed comparisons. ./src/H5Flow.c Fixed a warning about a variable being shadowed in the MPI-IO stuff. ./src/H5Iprivate.h ./src/H5Ipublic.h Added the H5_RAGGED atom group. ./src/H5Shyper.c Fixed some freeing-free-memory errors that resulted when certain arrays were freed but the pointers were left in the data structures. I simply set the pointers to null after they were freed. ./src/H5Sprivate.h ./src/H5Sselect.c Split the H5Sselect_hyperslab() function into an API and a private function so it could be called from the ragged array layer. Added H5S_SEL_ERROR and H5S_SEL_N to the switch statements to get rid or compiler warnings. ./src/H5Tconv.c Removed a misleading comment. ./test/bittests.c Fixed a warning about a printf(). ./test/cmpd_dset.c Fixed warnings about unused variables because of test #11 being commented out. ./bin/trace Shortened the right margin for the output to allow room for the `);' at the end of the TRACE() macros.
1998-08-28 00:48:50 +08:00
H5Gcreate(hid_t loc_id, const char *name, size_t size_hint)
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
{
[svn-r633] Changes since 19980828 ---------------------- ./RELEASE Updated with important changes I made since the second beta. ./src/H5A.c ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5R.c ./src/H5Rprivate.h ./src/H5T.c ./src/H5Tprivate.h Any API function that used to take an `hid_t loc_id' followed by a `const char *name' can now take any type of object for the loc_id as long as the object is somehow associated with a file. Internally, H5G_loc() was modified to return an H5G_entry_t* instead of an H5G_t* so it's more general. Among other things, this allows one to retrieve information about an object like a named type or dataset without knowing the name of the type or dataset: int get_nlinks (hid_t obj) { H5G_stat_t sb; if (H5Gstat(obj, ".", TRUE, &sb)<0) return -1; return sb.nlink; } ./test/gheap.c ./test/istore.c These files needed a couple of changes because they call some of the internal functions whose H5G_t arguments changed to H5G_entry_t. ./src/H5A.c Got rid of all the switch statements for getting symbol table entries for varous objects and replaced them with a call to H5G_loc() allowing attributes to automatically apply to any type of object that belongs to a file. ./test/Makefile.in Moved the ragged array tests from the normal list of tests to the `make timings' target. ./test/ragged.c Added rewrite tests -- rewrite the rows of a dataset changing the number of rows and the length of each row. ./test/mtime.c Added a test that checks that H5Gstat() can be called with a dataset as the first argument. ./src/H5S.c Added #ifdef HAVE_PARALLEL around code to check for the HDF5_MPI_OPT_TYPES environment variable because the global variable that gets set is #ifdef'd. ./bin/release bzip2 uses .bz2 as the file extension.
1998-08-31 21:46:47 +08:00
H5G_entry_t *loc = NULL;
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
H5G_t *grp = NULL;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
hid_t ret_value;
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_API(H5Gcreate, FAIL);
[svn-r429] Changes since 19980616 ---------------------- ./html/tracing.html NEW This entire update is to make it possible for the library to print the name, arguments, and return value of every API call without requiring any extra work from developers or app programmers. This file describes how this all works. ./configure.in Added the `--enable-tracing' switch. If you use it then the library will include code to print API function names, argument names and values, and function return values. However, you must then turn on the tracing by setting the HDF5_TRACE environment variable to a file descriptor number. The default is `--disable-tracing' since enabling it causes a slight increase in library size and a slowdown resulting from an extra function call for each API function call (I couldn't even measure the slowdown :-) ./bin/trace NEW A perl script that synchronizes the H5TRACE() macro calls in the *.c files with the function return type and formal argument names and types. If you use GNU make and gcc then this will be done automatically, otherwise just invoke this script with the names of one or more .c files. You could do it by hand to, but encoding argument types is a little tricky at first. ./config/commence.in Added the $(TRACE) macro, which defaults to the no-op. Added -D_POSIX_SOURCE to the compiler command line. ./src/Makefile.in Override the default for $(TRACE). ./config/depend.in Automatically calls $(TRACE) to synchronize the H5TRACE() macros in any source file that changed. As with makefile dependencies, one way to force synchronization of all files is to remove the `.depend' file. ./MANIFEST Added new files. ./src/H5Eprivate.h Modified HRETURN_ERROR() and HRETURN() for tracing. ./src/H5.c ./src/H5private.h This is where the real tracing work really happens, in H5_trace(). ./src/H5A.c ./src/H5D.c ./src/H5G.c ./src/H5P.c ./src/H5S.c ./src/H5Z.c Added H5TRACE() calls to all API functions. You don't really need these changes if you don't want to merge your stuff because they can be generated automatically by going to the hdf5/src directory and saying ../bin/trace *.c ./src/H5T.c Added H5TRACE() calls. Other stuff below. ./src/H5E.c ./src/H5Epublic.h Added H5TRACE() calls. Created a type H5E_auto_t for the `func' argument of H5Eset_auto() and H5Eget_auto() to make those arguments easier to parse for tracing. It should also make it clearer for users that don't know how to read complicated ANSI data types. ./src/H5F.c Added H5TRACE() calls. Changed a couple `uintn' argument types in API functions to `unsigned int' since `uintn' part of the API. Changed a few "can't" and "cant" error messages to "unable to". ./src/H5Ipublic.h Removed H5_DIRECTORY from the H5I_group_t enum. It wasn't used anywhere. ./src/H5Tconv.c Removed an unused label. ./src/H5Fistore.c ./src/H5Oattr.c ./src/H5Odtype.c ./src/H5T.c ./test/dsets.c ./test/dtypes.c Fixed a warning about a variable possibly used before it's initialized. Added __unused__ to turn off some unused argument warnings that pop up when debugging is turned off and optimizations are turned on.
1998-06-18 04:46:29 +08:00
H5TRACE3("i","isz",loc_id,name,size_hint);
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/* Check arguments */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (NULL==(loc=H5G_loc (loc_id)))
HGOTO_ERROR (H5E_ARGS, H5E_BADTYPE, FAIL, "not a location");
if (!name || !*name)
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "no name given");
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/* Create the group */
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (NULL == (grp = H5G_create(loc, name, size_hint, H5AC_dxpl_id)))
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, FAIL, "unable to create group");
if ((ret_value = H5I_register(H5I_GROUP, grp)) < 0)
HGOTO_ERROR(H5E_ATOM, H5E_CANTREGISTER, FAIL, "unable to register group");
done:
if(ret_value<0) {
if(grp!=NULL)
H5G_close(grp);
} /* end if */
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_API(ret_value);
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
}
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
/*-------------------------------------------------------------------------
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Function: H5Gopen
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Purpose: Opens an existing group for modification. When finished,
* call H5Gclose() to close it and release resources.
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
* Errors:
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Return: Success: Object ID of the group.
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Failure: FAIL
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Programmer: Robb Matzke
* Wednesday, December 31, 1997
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
hid_t
[svn-r620] Changes since 19980825 ---------------------- ./MANIFEST ./src/H5R.c [NEW] ./src/H5Rprivate.h [NEW] ./src/H5Rpublic.h [NEW] ./src/Makefile.in ./src/hdf5.h ./test/ragged.c [NEW] Preliminary support for 2d ragged arrays for Mark Miller and Jim Reus. Not fully implemented yet. The test is not actually part of `make test' because we still have some memory problems. ./src/H5E.c ./src/H5Epublic.h Added H5E_RAGGED as a major error number. ./bin/release Checks the MANIFEST file against `svf ls' on systems that have it. ./bin/trace Fixed a bug that caused arguments of type `void *x[]' to not be handled. ./src/H5.c Removed unused variables and changed a couple types to fix compiler warnings. Added tracing support for ragged array object ID's and arrays of pointers. ./src/H5D.c H5Dcreate() will complain if either of the property lists are invalid (instead of using the default). ./src/H5D.c ./src/H5Dprivate.h Split H5Dget_space() into an API and internal function so it can be called from the new ragged array layer. ./src/H5Fistore.c Fixed warnings about unsigned vs. signed comparisons. ./src/H5Flow.c Fixed a warning about a variable being shadowed in the MPI-IO stuff. ./src/H5Iprivate.h ./src/H5Ipublic.h Added the H5_RAGGED atom group. ./src/H5Shyper.c Fixed some freeing-free-memory errors that resulted when certain arrays were freed but the pointers were left in the data structures. I simply set the pointers to null after they were freed. ./src/H5Sprivate.h ./src/H5Sselect.c Split the H5Sselect_hyperslab() function into an API and a private function so it could be called from the ragged array layer. Added H5S_SEL_ERROR and H5S_SEL_N to the switch statements to get rid or compiler warnings. ./src/H5Tconv.c Removed a misleading comment. ./test/bittests.c Fixed a warning about a printf(). ./test/cmpd_dset.c Fixed warnings about unused variables because of test #11 being commented out. ./bin/trace Shortened the right margin for the output to allow room for the `);' at the end of the TRACE() macros.
1998-08-28 00:48:50 +08:00
H5Gopen(hid_t loc_id, const char *name)
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
{
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
hid_t ret_value = FAIL;
H5G_t *grp = NULL;
[svn-r633] Changes since 19980828 ---------------------- ./RELEASE Updated with important changes I made since the second beta. ./src/H5A.c ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5R.c ./src/H5Rprivate.h ./src/H5T.c ./src/H5Tprivate.h Any API function that used to take an `hid_t loc_id' followed by a `const char *name' can now take any type of object for the loc_id as long as the object is somehow associated with a file. Internally, H5G_loc() was modified to return an H5G_entry_t* instead of an H5G_t* so it's more general. Among other things, this allows one to retrieve information about an object like a named type or dataset without knowing the name of the type or dataset: int get_nlinks (hid_t obj) { H5G_stat_t sb; if (H5Gstat(obj, ".", TRUE, &sb)<0) return -1; return sb.nlink; } ./test/gheap.c ./test/istore.c These files needed a couple of changes because they call some of the internal functions whose H5G_t arguments changed to H5G_entry_t. ./src/H5A.c Got rid of all the switch statements for getting symbol table entries for varous objects and replaced them with a call to H5G_loc() allowing attributes to automatically apply to any type of object that belongs to a file. ./test/Makefile.in Moved the ragged array tests from the normal list of tests to the `make timings' target. ./test/ragged.c Added rewrite tests -- rewrite the rows of a dataset changing the number of rows and the length of each row. ./test/mtime.c Added a test that checks that H5Gstat() can be called with a dataset as the first argument. ./src/H5S.c Added #ifdef HAVE_PARALLEL around code to check for the HDF5_MPI_OPT_TYPES environment variable because the global variable that gets set is #ifdef'd. ./bin/release bzip2 uses .bz2 as the file extension.
1998-08-31 21:46:47 +08:00
H5G_entry_t *loc = NULL;
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_API(H5Gopen, FAIL);
[svn-r429] Changes since 19980616 ---------------------- ./html/tracing.html NEW This entire update is to make it possible for the library to print the name, arguments, and return value of every API call without requiring any extra work from developers or app programmers. This file describes how this all works. ./configure.in Added the `--enable-tracing' switch. If you use it then the library will include code to print API function names, argument names and values, and function return values. However, you must then turn on the tracing by setting the HDF5_TRACE environment variable to a file descriptor number. The default is `--disable-tracing' since enabling it causes a slight increase in library size and a slowdown resulting from an extra function call for each API function call (I couldn't even measure the slowdown :-) ./bin/trace NEW A perl script that synchronizes the H5TRACE() macro calls in the *.c files with the function return type and formal argument names and types. If you use GNU make and gcc then this will be done automatically, otherwise just invoke this script with the names of one or more .c files. You could do it by hand to, but encoding argument types is a little tricky at first. ./config/commence.in Added the $(TRACE) macro, which defaults to the no-op. Added -D_POSIX_SOURCE to the compiler command line. ./src/Makefile.in Override the default for $(TRACE). ./config/depend.in Automatically calls $(TRACE) to synchronize the H5TRACE() macros in any source file that changed. As with makefile dependencies, one way to force synchronization of all files is to remove the `.depend' file. ./MANIFEST Added new files. ./src/H5Eprivate.h Modified HRETURN_ERROR() and HRETURN() for tracing. ./src/H5.c ./src/H5private.h This is where the real tracing work really happens, in H5_trace(). ./src/H5A.c ./src/H5D.c ./src/H5G.c ./src/H5P.c ./src/H5S.c ./src/H5Z.c Added H5TRACE() calls to all API functions. You don't really need these changes if you don't want to merge your stuff because they can be generated automatically by going to the hdf5/src directory and saying ../bin/trace *.c ./src/H5T.c Added H5TRACE() calls. Other stuff below. ./src/H5E.c ./src/H5Epublic.h Added H5TRACE() calls. Created a type H5E_auto_t for the `func' argument of H5Eset_auto() and H5Eget_auto() to make those arguments easier to parse for tracing. It should also make it clearer for users that don't know how to read complicated ANSI data types. ./src/H5F.c Added H5TRACE() calls. Changed a couple `uintn' argument types in API functions to `unsigned int' since `uintn' part of the API. Changed a few "can't" and "cant" error messages to "unable to". ./src/H5Ipublic.h Removed H5_DIRECTORY from the H5I_group_t enum. It wasn't used anywhere. ./src/H5Tconv.c Removed an unused label. ./src/H5Fistore.c ./src/H5Oattr.c ./src/H5Odtype.c ./src/H5T.c ./test/dsets.c ./test/dtypes.c Fixed a warning about a variable possibly used before it's initialized. Added __unused__ to turn off some unused argument warnings that pop up when debugging is turned off and optimizations are turned on.
1998-06-18 04:46:29 +08:00
H5TRACE2("i","is",loc_id,name);
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/* Check args */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (NULL==(loc=H5G_loc(loc_id)))
HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, FAIL, "not a location");
if (!name || !*name)
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "no name");
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/* Open the group */
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (NULL == (grp = H5G_open(loc, name, H5AC_dxpl_id)))
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTOPENOBJ, FAIL, "unable to open group");
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/* Register an atom for the group */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if ((ret_value = H5I_register(H5I_GROUP, grp)) < 0)
HGOTO_ERROR(H5E_ATOM, H5E_CANTREGISTER, FAIL, "unable to register group");
done:
if(ret_value<0) {
if(grp!=NULL)
H5G_close(grp);
} /* end if */
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_API(ret_value);
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
}
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
/*-------------------------------------------------------------------------
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Function: H5Gclose
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Purpose: Closes the specified group. The group ID will no longer be
* valid for accessing the group.
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Return: Non-negative on success/Negative on failure
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Programmer: Robb Matzke
* Wednesday, December 31, 1997
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
herr_t
[svn-r620] Changes since 19980825 ---------------------- ./MANIFEST ./src/H5R.c [NEW] ./src/H5Rprivate.h [NEW] ./src/H5Rpublic.h [NEW] ./src/Makefile.in ./src/hdf5.h ./test/ragged.c [NEW] Preliminary support for 2d ragged arrays for Mark Miller and Jim Reus. Not fully implemented yet. The test is not actually part of `make test' because we still have some memory problems. ./src/H5E.c ./src/H5Epublic.h Added H5E_RAGGED as a major error number. ./bin/release Checks the MANIFEST file against `svf ls' on systems that have it. ./bin/trace Fixed a bug that caused arguments of type `void *x[]' to not be handled. ./src/H5.c Removed unused variables and changed a couple types to fix compiler warnings. Added tracing support for ragged array object ID's and arrays of pointers. ./src/H5D.c H5Dcreate() will complain if either of the property lists are invalid (instead of using the default). ./src/H5D.c ./src/H5Dprivate.h Split H5Dget_space() into an API and internal function so it can be called from the new ragged array layer. ./src/H5Fistore.c Fixed warnings about unsigned vs. signed comparisons. ./src/H5Flow.c Fixed a warning about a variable being shadowed in the MPI-IO stuff. ./src/H5Iprivate.h ./src/H5Ipublic.h Added the H5_RAGGED atom group. ./src/H5Shyper.c Fixed some freeing-free-memory errors that resulted when certain arrays were freed but the pointers were left in the data structures. I simply set the pointers to null after they were freed. ./src/H5Sprivate.h ./src/H5Sselect.c Split the H5Sselect_hyperslab() function into an API and a private function so it could be called from the ragged array layer. Added H5S_SEL_ERROR and H5S_SEL_N to the switch statements to get rid or compiler warnings. ./src/H5Tconv.c Removed a misleading comment. ./test/bittests.c Fixed a warning about a printf(). ./test/cmpd_dset.c Fixed warnings about unused variables because of test #11 being commented out. ./bin/trace Shortened the right margin for the output to allow room for the `);' at the end of the TRACE() macros.
1998-08-28 00:48:50 +08:00
H5Gclose(hid_t group_id)
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
{
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_API(H5Gclose, FAIL);
[svn-r529] Changes since 19980722 ---------------------- ./src/H5.c Handle hid_t of type H5_TEMPBUF, arguments usually called tbuf_id. Added array tracing where the array rank is stored in a simple data space. Just use the name of the data space argument when declaring the array argument: herr_t H5Sselect_hyperslab (hid_t space_id, H5S_seloper_t op, const hssize_t start[/*space_id*/], const hsize_t _stride[/*space_id*/], const hsize_t count[/*space_id*/], const hsize_t _block[/*space_id*/]) and when the program runs you'll see array values printed: H5Sselect_hyperslab(space=218103813, op=H5S_SELECT_SET, start=0xbfffef4c {0}, _stride=NULL, count=0xbfffef44 {64}, _block=NULL) = SUCCEED; Added more symbolic data types to the tracing output. ./src/H5A.c ./src/H5Apublic.h ./src/H5D.c ./src/H5Dpublic.h ./src/H5F.c ./src/H5Fpublic.h ./src/H5G.c ./src/H5Gpublic.h ./src/H5P.c ./src/H5Ppublic.h ./src/H5S.c ./src/H5Sall.c ./src/H5Shyper.c ./src/H5Spoint.c ./src/H5Spublic.h ./src/H5Sselect.c ./src/H5Ssimp.c ./src/H5TB.c ./src/H5V.c Changed some API argument names to be more consistent with other API functions and to produce better tracing output. Reformatted some long lines. Indented printf statements. ./tools/h5ls.c Fixed warnings about unsigned vs. signed comparisons.
1998-07-23 02:45:46 +08:00
H5TRACE1("e","i",group_id);
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/* Check args */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (NULL == H5I_object_verify(group_id,H5I_GROUP))
HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, FAIL, "not a group");
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Decrement the counter on the group atom. It will be freed if the count
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
* reaches zero.
*/
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (H5I_dec_ref(group_id) < 0)
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, FAIL, "unable to close group");
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_API(ret_value);
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
}
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r328] Changes since 19980318 ---------------------- ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gpublic.h ./src/H5Gstab.c Implemented H5Giterate(). However, since most functions can't take a group ID in place of a file ID yet, there's not a whole lot that the operator can do besides print the name or something. ./test/H5O.c Fixed writing of four uninitialized bytes to the file as part of an object header. ./test/istore.c For some reason, `mpirun -np 1 istore' results in extra arguments on the command line that istore doesn't understand. I'm probably forgetting to call some MPI function in main(). Albert, Kim? So I commented out the `exit(1)' for the time being.
1998-03-24 22:53:57 +08:00
/*-------------------------------------------------------------------------
* Function: H5Giterate
*
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
* Purpose: Iterates over the entries of a group. The LOC_ID and NAME
* identify the group over which to iterate and IDX indicates
[svn-r328] Changes since 19980318 ---------------------- ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gpublic.h ./src/H5Gstab.c Implemented H5Giterate(). However, since most functions can't take a group ID in place of a file ID yet, there's not a whole lot that the operator can do besides print the name or something. ./test/H5O.c Fixed writing of four uninitialized bytes to the file as part of an object header. ./test/istore.c For some reason, `mpirun -np 1 istore' results in extra arguments on the command line that istore doesn't understand. I'm probably forgetting to call some MPI function in main(). Albert, Kim? So I commented out the `exit(1)' for the time being.
1998-03-24 22:53:57 +08:00
* where to start iterating (zero means at the beginning). The
* OPERATOR is called for each member and the iteration
* continues until the operator returns non-zero or all members
* are processed. The operator is passed a group ID for the
* group being iterated, a member name, and OP_DATA for each
* member.
*
* Return: Success: The return value of the first operator that
* returns non-zero, or zero if all members were
* processed with no operator returning non-zero.
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Failure: Negative if something goes wrong within the
* library, or the negative value returned by one
[svn-r328] Changes since 19980318 ---------------------- ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gpublic.h ./src/H5Gstab.c Implemented H5Giterate(). However, since most functions can't take a group ID in place of a file ID yet, there's not a whole lot that the operator can do besides print the name or something. ./test/H5O.c Fixed writing of four uninitialized bytes to the file as part of an object header. ./test/istore.c For some reason, `mpirun -np 1 istore' results in extra arguments on the command line that istore doesn't understand. I'm probably forgetting to call some MPI function in main(). Albert, Kim? So I commented out the `exit(1)' for the time being.
1998-03-24 22:53:57 +08:00
* of the operators.
*
* Programmer: Robb Matzke
* Monday, March 23, 1998
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
herr_t
[svn-r7915] Purpose: Bug fix Description: Make a copy of the index value for H5Giterate and use that instead of dereferencing the index pointer. Platforms tested: FreeBSD 4.9 (sleipnir) too minor to need h5committest
2003-12-06 22:57:06 +08:00
H5Giterate(hid_t loc_id, const char *name, int *idx_p,
[svn-r328] Changes since 19980318 ---------------------- ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gpublic.h ./src/H5Gstab.c Implemented H5Giterate(). However, since most functions can't take a group ID in place of a file ID yet, there's not a whole lot that the operator can do besides print the name or something. ./test/H5O.c Fixed writing of four uninitialized bytes to the file as part of an object header. ./test/istore.c For some reason, `mpirun -np 1 istore' results in extra arguments on the command line that istore doesn't understand. I'm probably forgetting to call some MPI function in main(). Albert, Kim? So I commented out the `exit(1)' for the time being.
1998-03-24 22:53:57 +08:00
H5G_iterate_t op, void *op_data)
{
[svn-r7915] Purpose: Bug fix Description: Make a copy of the index value for H5Giterate and use that instead of dereferencing the index pointer. Platforms tested: FreeBSD 4.9 (sleipnir) too minor to need h5committest
2003-12-06 22:57:06 +08:00
int idx;
[svn-r328] Changes since 19980318 ---------------------- ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gpublic.h ./src/H5Gstab.c Implemented H5Giterate(). However, since most functions can't take a group ID in place of a file ID yet, there's not a whole lot that the operator can do besides print the name or something. ./test/H5O.c Fixed writing of four uninitialized bytes to the file as part of an object header. ./test/istore.c For some reason, `mpirun -np 1 istore' results in extra arguments on the command line that istore doesn't understand. I'm probably forgetting to call some MPI function in main(). Albert, Kim? So I commented out the `exit(1)' for the time being.
1998-03-24 22:53:57 +08:00
H5G_bt_ud2_t udata;
[svn-r633] Changes since 19980828 ---------------------- ./RELEASE Updated with important changes I made since the second beta. ./src/H5A.c ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5R.c ./src/H5Rprivate.h ./src/H5T.c ./src/H5Tprivate.h Any API function that used to take an `hid_t loc_id' followed by a `const char *name' can now take any type of object for the loc_id as long as the object is somehow associated with a file. Internally, H5G_loc() was modified to return an H5G_entry_t* instead of an H5G_t* so it's more general. Among other things, this allows one to retrieve information about an object like a named type or dataset without knowing the name of the type or dataset: int get_nlinks (hid_t obj) { H5G_stat_t sb; if (H5Gstat(obj, ".", TRUE, &sb)<0) return -1; return sb.nlink; } ./test/gheap.c ./test/istore.c These files needed a couple of changes because they call some of the internal functions whose H5G_t arguments changed to H5G_entry_t. ./src/H5A.c Got rid of all the switch statements for getting symbol table entries for varous objects and replaced them with a call to H5G_loc() allowing attributes to automatically apply to any type of object that belongs to a file. ./test/Makefile.in Moved the ragged array tests from the normal list of tests to the `make timings' target. ./test/ragged.c Added rewrite tests -- rewrite the rows of a dataset changing the number of rows and the length of each row. ./test/mtime.c Added a test that checks that H5Gstat() can be called with a dataset as the first argument. ./src/H5S.c Added #ifdef HAVE_PARALLEL around code to check for the HDF5_MPI_OPT_TYPES environment variable because the global variable that gets set is #ifdef'd. ./bin/release bzip2 uses .bz2 as the file extension.
1998-08-31 21:46:47 +08:00
H5G_entry_t *loc = NULL;
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
H5G_t *grp = NULL;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value;
[svn-r328] Changes since 19980318 ---------------------- ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gpublic.h ./src/H5Gstab.c Implemented H5Giterate(). However, since most functions can't take a group ID in place of a file ID yet, there's not a whole lot that the operator can do besides print the name or something. ./test/H5O.c Fixed writing of four uninitialized bytes to the file as part of an object header. ./test/istore.c For some reason, `mpirun -np 1 istore' results in extra arguments on the command line that istore doesn't understand. I'm probably forgetting to call some MPI function in main(). Albert, Kim? So I commented out the `exit(1)' for the time being.
1998-03-24 22:53:57 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_API(H5Giterate, FAIL);
[svn-r7915] Purpose: Bug fix Description: Make a copy of the index value for H5Giterate and use that instead of dereferencing the index pointer. Platforms tested: FreeBSD 4.9 (sleipnir) too minor to need h5committest
2003-12-06 22:57:06 +08:00
H5TRACE5("e","is*Isxx",loc_id,name,idx_p,op,op_data);
[svn-r328] Changes since 19980318 ---------------------- ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gpublic.h ./src/H5Gstab.c Implemented H5Giterate(). However, since most functions can't take a group ID in place of a file ID yet, there's not a whole lot that the operator can do besides print the name or something. ./test/H5O.c Fixed writing of four uninitialized bytes to the file as part of an object header. ./test/istore.c For some reason, `mpirun -np 1 istore' results in extra arguments on the command line that istore doesn't understand. I'm probably forgetting to call some MPI function in main(). Albert, Kim? So I commented out the `exit(1)' for the time being.
1998-03-24 22:53:57 +08:00
/* Check args */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (NULL==(loc=H5G_loc (loc_id)))
HGOTO_ERROR (H5E_ARGS, H5E_BADTYPE, FAIL, "not a location");
if (!name || !*name)
HGOTO_ERROR (H5E_ARGS, H5E_BADVALUE, FAIL, "no name specified");
[svn-r7915] Purpose: Bug fix Description: Make a copy of the index value for H5Giterate and use that instead of dereferencing the index pointer. Platforms tested: FreeBSD 4.9 (sleipnir) too minor to need h5committest
2003-12-06 22:57:06 +08:00
idx = (idx_p == NULL ? 0 : *idx_p);
if (!idx_p)
idx_p = &idx;
if (idx<0)
[svn-r7082] Purpose: Bug fix (backward compatibility) Description: Track changes to allow H5Giterate to pass along iterator callback's return value Also, improve range checking on "index" parameter to disallow invalid starting indices (<0 or > the number of objects in a group). Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-06-23 21:27:02 +08:00
HGOTO_ERROR (H5E_ARGS, H5E_BADVALUE, FAIL, "invalid index specified");
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (!op)
HGOTO_ERROR (H5E_ARGS, H5E_BADVALUE, FAIL, "no operator specified");
[svn-r328] Changes since 19980318 ---------------------- ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gpublic.h ./src/H5Gstab.c Implemented H5Giterate(). However, since most functions can't take a group ID in place of a file ID yet, there's not a whole lot that the operator can do besides print the name or something. ./test/H5O.c Fixed writing of four uninitialized bytes to the file as part of an object header. ./test/istore.c For some reason, `mpirun -np 1 istore' results in extra arguments on the command line that istore doesn't understand. I'm probably forgetting to call some MPI function in main(). Albert, Kim? So I commented out the `exit(1)' for the time being.
1998-03-24 22:53:57 +08:00
/*
* Open the group on which to operate. We also create a group ID which
* we can pass to the application-defined operator.
*/
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
if (NULL==(grp = H5G_open (loc, name, H5AC_dxpl_id)))
[svn-r7082] Purpose: Bug fix (backward compatibility) Description: Track changes to allow H5Giterate to pass along iterator callback's return value Also, improve range checking on "index" parameter to disallow invalid starting indices (<0 or > the number of objects in a group). Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-06-23 21:27:02 +08:00
HGOTO_ERROR (H5E_SYM, H5E_CANTOPENOBJ, FAIL, "unable to open group");
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
if ((udata.group_id=H5I_register (H5I_GROUP, grp))<0) {
H5G_close(grp);
[svn-r7082] Purpose: Bug fix (backward compatibility) Description: Track changes to allow H5Giterate to pass along iterator callback's return value Also, improve range checking on "index" parameter to disallow invalid starting indices (<0 or > the number of objects in a group). Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-06-23 21:27:02 +08:00
HGOTO_ERROR (H5E_SYM, H5E_CANTREGISTER, FAIL, "unable to register group");
[svn-r328] Changes since 19980318 ---------------------- ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gpublic.h ./src/H5Gstab.c Implemented H5Giterate(). However, since most functions can't take a group ID in place of a file ID yet, there's not a whole lot that the operator can do besides print the name or something. ./test/H5O.c Fixed writing of four uninitialized bytes to the file as part of an object header. ./test/istore.c For some reason, `mpirun -np 1 istore' results in extra arguments on the command line that istore doesn't understand. I'm probably forgetting to call some MPI function in main(). Albert, Kim? So I commented out the `exit(1)' for the time being.
1998-03-24 22:53:57 +08:00
}
/* Build udata to pass through H5B_iterate() to H5G_node_iterate() */
[svn-r7915] Purpose: Bug fix Description: Make a copy of the index value for H5Giterate and use that instead of dereferencing the index pointer. Platforms tested: FreeBSD 4.9 (sleipnir) too minor to need h5committest
2003-12-06 22:57:06 +08:00
udata.skip = idx;
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
udata.ent = &(grp->ent);
[svn-r328] Changes since 19980318 ---------------------- ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gpublic.h ./src/H5Gstab.c Implemented H5Giterate(). However, since most functions can't take a group ID in place of a file ID yet, there's not a whole lot that the operator can do besides print the name or something. ./test/H5O.c Fixed writing of four uninitialized bytes to the file as part of an object header. ./test/istore.c For some reason, `mpirun -np 1 istore' results in extra arguments on the command line that istore doesn't understand. I'm probably forgetting to call some MPI function in main(). Albert, Kim? So I commented out the `exit(1)' for the time being.
1998-03-24 22:53:57 +08:00
udata.op = op;
udata.op_data = op_data;
[svn-r2081] Fixed a problem in H5Giterate which was not updating the 'index' paramater for certain return values from the callback.
2000-04-06 04:51:27 +08:00
/* Set the number of entries looked at to zero */
udata.final_ent = 0;
[svn-r328] Changes since 19980318 ---------------------- ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gpublic.h ./src/H5Gstab.c Implemented H5Giterate(). However, since most functions can't take a group ID in place of a file ID yet, there's not a whole lot that the operator can do besides print the name or something. ./test/H5O.c Fixed writing of four uninitialized bytes to the file as part of an object header. ./test/istore.c For some reason, `mpirun -np 1 istore' results in extra arguments on the command line that istore doesn't understand. I'm probably forgetting to call some MPI function in main(). Albert, Kim? So I commented out the `exit(1)' for the time being.
1998-03-24 22:53:57 +08:00
/* Iterate over the group members */
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
if ((ret_value = H5B_iterate (H5G_fileof(grp), H5AC_dxpl_id, H5B_SNODE,
H5G_node_iterate, udata.ent->cache.stab.btree_addr, &udata))<0)
[svn-r7082] Purpose: Bug fix (backward compatibility) Description: Track changes to allow H5Giterate to pass along iterator callback's return value Also, improve range checking on "index" parameter to disallow invalid starting indices (<0 or > the number of objects in a group). Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-06-23 21:27:02 +08:00
HERROR (H5E_SYM, H5E_CANTNEXT, "iteration operator failed");
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
H5I_dec_ref (udata.group_id); /*also closes 'grp'*/
[svn-r7082] Purpose: Bug fix (backward compatibility) Description: Track changes to allow H5Giterate to pass along iterator callback's return value Also, improve range checking on "index" parameter to disallow invalid starting indices (<0 or > the number of objects in a group). Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-06-23 21:27:02 +08:00
/* Check for too high of a starting index (ex post facto :-) */
/* (Skipping exactly as many entries as are in the group is currently an error) */
[svn-r7915] Purpose: Bug fix Description: Make a copy of the index value for H5Giterate and use that instead of dereferencing the index pointer. Platforms tested: FreeBSD 4.9 (sleipnir) too minor to need h5committest
2003-12-06 22:57:06 +08:00
if (idx>0 && idx>=udata.final_ent)
[svn-r7082] Purpose: Bug fix (backward compatibility) Description: Track changes to allow H5Giterate to pass along iterator callback's return value Also, improve range checking on "index" parameter to disallow invalid starting indices (<0 or > the number of objects in a group). Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-06-23 21:27:02 +08:00
HGOTO_ERROR (H5E_ARGS, H5E_BADVALUE, FAIL, "invalid index specified");
[svn-r2081] Fixed a problem in H5Giterate which was not updating the 'index' paramater for certain return values from the callback.
2000-04-06 04:51:27 +08:00
/* Set the index we stopped at */
[svn-r7915] Purpose: Bug fix Description: Make a copy of the index value for H5Giterate and use that instead of dereferencing the index pointer. Platforms tested: FreeBSD 4.9 (sleipnir) too minor to need h5committest
2003-12-06 22:57:06 +08:00
*idx_p=udata.final_ent;
[svn-r2081] Fixed a problem in H5Giterate which was not updating the 'index' paramater for certain return values from the callback.
2000-04-06 04:51:27 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_API(ret_value);
[svn-r328] Changes since 19980318 ---------------------- ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gpublic.h ./src/H5Gstab.c Implemented H5Giterate(). However, since most functions can't take a group ID in place of a file ID yet, there's not a whole lot that the operator can do besides print the name or something. ./test/H5O.c Fixed writing of four uninitialized bytes to the file as part of an object header. ./test/istore.c For some reason, `mpirun -np 1 istore' results in extra arguments on the command line that istore doesn't understand. I'm probably forgetting to call some MPI function in main(). Albert, Kim? So I commented out the `exit(1)' for the time being.
1998-03-24 22:53:57 +08:00
}
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
/*-------------------------------------------------------------------------
* Function: H5Gget_num_objs
*
* Purpose: Returns the number of objects in the group. It iterates
* all B-tree leaves and sum up total number of group members.
*
* Return: Success: Non-negative
*
* Failure: Negative
*
* Programmer: Raymond Lu
* Nov 20, 2002
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
herr_t
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
H5Gget_num_objs(hid_t loc_id, hsize_t *num_objs)
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
{
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
H5G_entry_t *loc = NULL; /* Pointer to symbol table entry */
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
herr_t ret_value;
FUNC_ENTER_API(H5Gget_num_objs, FAIL);
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
H5TRACE2("e","i*h",loc_id,num_objs);
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
/* Check args */
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
if (NULL==(loc=H5G_loc (loc_id)))
HGOTO_ERROR (H5E_ARGS, H5E_BADTYPE, FAIL, "not a location ID");
if(H5G_get_type(loc,H5AC_ind_dxpl_id)!=H5G_GROUP)
HGOTO_ERROR (H5E_ARGS, H5E_BADTYPE, FAIL, "not a group");
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
if (!num_objs)
HGOTO_ERROR (H5E_ARGS, H5E_BADVALUE, FAIL, "nil pointer");
/* Call private function. */
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
ret_value = H5G_get_num_objs(loc, num_objs, H5AC_ind_dxpl_id);
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_API(ret_value);
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
}
/*-------------------------------------------------------------------------
* Function: H5Gget_objname_by_idx
*
* Purpose: Returns the name of objects in the group by giving index.
* If `name' is non-NULL then write up to `size' bytes into that
* buffer and always return the length of the entry name.
* Otherwise `size' is ignored and the function does not store the name,
* just returning the number of characters required to store the name.
* If an error occurs then the buffer pointed to by `name' (NULL or non-NULL)
* is unchanged and the function returns a negative value.
* If a zero is returned for the name's length, then there is no name
* associated with the ID.
*
* Return: Success: Non-negative
*
* Failure: Negative
*
* Programmer: Raymond Lu
* Nov 20, 2002
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
ssize_t
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
H5Gget_objname_by_idx(hid_t loc_id, hsize_t idx, char *name, size_t size)
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
{
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
H5G_entry_t *loc = NULL; /* Pointer to symbol table entry */
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
hsize_t num_objs;
ssize_t ret_value = FAIL;
FUNC_ENTER_API(H5Gget_objname_by_idx, FAIL);
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
H5TRACE4("Zs","ihsz",loc_id,idx,name,size);
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
/* Check args */
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
if (NULL==(loc=H5G_loc (loc_id)))
HGOTO_ERROR (H5E_ARGS, H5E_BADTYPE, FAIL, "not a location ID");
if(H5G_get_type(loc,H5AC_ind_dxpl_id)!=H5G_GROUP)
HGOTO_ERROR (H5E_ARGS, H5E_BADTYPE, FAIL, "not a group");
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
if (H5G_get_num_objs(loc, &num_objs, H5AC_ind_dxpl_id)<0)
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, FAIL, "unable to retrieve number of members");
if(idx >= num_objs)
HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, FAIL, "index out of bound");
/*call private function*/
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
ret_value = H5G_get_objname_by_idx(loc, idx, name, size, H5AC_ind_dxpl_id);
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_API(ret_value);
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
}
/*-------------------------------------------------------------------------
* Function: H5Gget_objtype_by_idx
*
* Purpose: Returns the type of objects in the group by giving index.
*
*
* Return: Success: H5G_GROUP(1), H5G_DATASET(2), H5G_TYPE(3)
*
* Failure: H5G_UNKNOWN
*
* Programmer: Raymond Lu
* Nov 20, 2002
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
[svn-r6744] Purpose: A little code rewriting Description: object types were defined as macros in H5Gpublic.h Solution: changed them to enumerate type Platforms tested: h5committtest
2003-04-25 04:15:42 +08:00
H5G_obj_t
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
H5Gget_objtype_by_idx(hid_t loc_id, hsize_t idx)
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
{
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
H5G_entry_t *loc = NULL; /* Pointer to symbol table entry */
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
hsize_t num_objs;
[svn-r6843] Purpose: Code cleanup Description: Clean up warnings exposed by compiling on O2K. Also, revert some of Bill and my changes to the H5S_mpi_opt_types_g, etc. and settle them back into their original location. Platforms tested: h5committested.
2003-05-10 02:18:21 +08:00
H5G_obj_t ret_value;
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
[svn-r6843] Purpose: Code cleanup Description: Clean up warnings exposed by compiling on O2K. Also, revert some of Bill and my changes to the H5S_mpi_opt_types_g, etc. and settle them back into their original location. Platforms tested: h5committested.
2003-05-10 02:18:21 +08:00
FUNC_ENTER_API(H5Gget_objtype_by_idx, H5G_UNKNOWN);
[svn-r7626] Purpose: Update dependencies and tracing information Platforms tested: Linux 2.4 (verbena) w/FORTRAN too minor for h5committest
2003-10-14 23:46:38 +08:00
H5TRACE2("Go","ih",loc_id,idx);
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
/* Check args */
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
if (NULL==(loc=H5G_loc (loc_id)))
[svn-r7638] Purpose: Code cleanup Description: Clean up compiler warnings Platforms tested: h5committest
2003-10-16 04:02:51 +08:00
HGOTO_ERROR (H5E_ARGS, H5E_BADTYPE, H5G_UNKNOWN, "not a location ID");
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
if(H5G_get_type(loc,H5AC_ind_dxpl_id)!=H5G_GROUP)
[svn-r7638] Purpose: Code cleanup Description: Clean up compiler warnings Platforms tested: h5committest
2003-10-16 04:02:51 +08:00
HGOTO_ERROR (H5E_ARGS, H5E_BADTYPE, H5G_UNKNOWN, "not a group");
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
if (H5G_get_num_objs(loc, &num_objs, H5AC_ind_dxpl_id)<0)
[svn-r6843] Purpose: Code cleanup Description: Clean up warnings exposed by compiling on O2K. Also, revert some of Bill and my changes to the H5S_mpi_opt_types_g, etc. and settle them back into their original location. Platforms tested: h5committested.
2003-05-10 02:18:21 +08:00
HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, H5G_UNKNOWN, "unable to retrieve number of members");
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
if(idx >= num_objs)
[svn-r6843] Purpose: Code cleanup Description: Clean up warnings exposed by compiling on O2K. Also, revert some of Bill and my changes to the H5S_mpi_opt_types_g, etc. and settle them back into their original location. Platforms tested: h5committested.
2003-05-10 02:18:21 +08:00
HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, H5G_UNKNOWN, "index out of bound");
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
/*call private function*/
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
ret_value = H5G_get_objtype_by_idx(loc, idx, H5AC_ind_dxpl_id);
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_API(ret_value);
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
}
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
/*-------------------------------------------------------------------------
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
* Function: H5Gmove2
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
*
* Purpose: Renames an object within an HDF5 file. The original name SRC
* is unlinked from the group graph and the new name DST is
* inserted as an atomic operation. Both names are interpreted
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
* relative to SRC_LOC_ID and DST_LOC_ID, which are either a file
* ID or a group ID.
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Return: Non-negative on success/Negative on failure
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
*
* Programmer: Robb Matzke
* Monday, April 6, 1998
*
* Modifications:
*
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
* Raymond Lu
* Thursday, April 18, 2002
*
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
*-------------------------------------------------------------------------
*/
herr_t
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
H5Gmove2(hid_t src_loc_id, const char *src_name, hid_t dst_loc_id,
const char *dst_name)
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
{
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
H5G_entry_t *src_loc=NULL;
H5G_entry_t *dst_loc=NULL;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_API(H5Gmove2, FAIL);
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
H5TRACE4("e","isis",src_loc_id,src_name,dst_loc_id,dst_name);
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (src_loc_id != H5G_SAME_LOC && NULL==(src_loc=H5G_loc(src_loc_id)))
HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, FAIL, "not a location");
if (dst_loc_id != H5G_SAME_LOC && NULL==(dst_loc=H5G_loc(dst_loc_id)))
HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, FAIL, "not a location");
if (!src_name || !*src_name)
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "no current name specified");
if (!dst_name || !*dst_name)
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "no new name specified");
[svn-r726] Changes since 19980924 ---------------------- ./MANIFEST ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./test/Makefile.in ./test/unlink.c [NEW] Finished H5Gunlink() and H5Grename(). ./src/H5F.c ./src/H5Fistore.c ./src/H5Fprivate.h Removed the last memcpy() from the chunk cache. ./src/H5Fistore.c The offset of a chunk within a dataset is an 8-byte quantity per dimension instead of 4 bytes. ./src/H5HL.c Fixed infinite loops in H5HL_remove().
1998-09-28 22:20:21 +08:00
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
if(src_loc_id == H5G_SAME_LOC && dst_loc_id == H5G_SAME_LOC) {
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "source and destination should not be both H5G_SAME_LOC");
} else if(src_loc_id == H5G_SAME_LOC) {
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
src_loc = dst_loc;
}
else if(dst_loc_id == H5G_SAME_LOC) {
dst_loc = src_loc;
}
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
else if(src_loc->file != dst_loc->file)
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "source and destination should be in the same file.");
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_move(src_loc, src_name, dst_loc, dst_name, H5AC_dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, FAIL, "unable to change object name");
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_API(ret_value);
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
}
/*-------------------------------------------------------------------------
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
* Function: H5Glink2
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
*
* Purpose: Creates a link of the specified type from NEW_NAME to
* CUR_NAME.
*
* If TYPE is H5G_LINK_HARD then CUR_NAME must name an existing
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
* object. CUR_NAME and NEW_NAME are interpreted relative to
* CUR_LOC_ID and NEW_LOC_ID, which is either a file ID or a
* group ID.
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
*
* If TYPE is H5G_LINK_SOFT then CUR_NAME can be anything and is
* interpreted at lookup time relative to the group which
* contains the final component of NEW_NAME. For instance, if
* CUR_NAME is `./foo' and NEW_NAME is `./x/y/bar' and a request
* is made for `./x/y/bar' then the actual object looked up is
* `./x/y/./foo'.
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Return: Non-negative on success/Negative on failure
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
*
* Programmer: Robb Matzke
* Monday, April 6, 1998
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
herr_t
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
H5Glink2(hid_t cur_loc_id, const char *cur_name, H5G_link_t type,
hid_t new_loc_id, const char *new_name)
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
{
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
H5G_entry_t *cur_loc = NULL;
H5G_entry_t *new_loc = NULL;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_API(H5Glink2, FAIL);
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
H5TRACE5("e","isGlis",cur_loc_id,cur_name,type,new_loc_id,new_name);
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/* Check arguments */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (cur_loc_id != H5G_SAME_LOC && NULL==(cur_loc=H5G_loc(cur_loc_id)))
HGOTO_ERROR (H5E_ARGS, H5E_BADTYPE, FAIL, "not a location");
if (new_loc_id != H5G_SAME_LOC && NULL==(new_loc=H5G_loc(new_loc_id)))
HGOTO_ERROR (H5E_ARGS, H5E_BADTYPE, FAIL, "not a location");
if (type!=H5G_LINK_HARD && type!=H5G_LINK_SOFT)
HGOTO_ERROR (H5E_ARGS, H5E_BADVALUE, FAIL, "unrecognized link type");
if (!cur_name || !*cur_name)
HGOTO_ERROR (H5E_ARGS, H5E_BADVALUE, FAIL, "no current name specified");
if (!new_name || !*new_name)
HGOTO_ERROR (H5E_ARGS, H5E_BADVALUE, FAIL, "no new name specified");
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
if(cur_loc_id == H5G_SAME_LOC && new_loc_id == H5G_SAME_LOC) {
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "source and destination should not be both H5G_SAME_LOC");
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
}
else if(cur_loc_id == H5G_SAME_LOC) {
cur_loc = new_loc;
}
else if(new_loc_id == H5G_SAME_LOC) {
new_loc = cur_loc;
}
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
else if(cur_loc->file != new_loc->file)
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "source and destination should be in the same file.");
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_link(cur_loc, cur_name, new_loc, new_name, type, H5G_TARGET_NORMAL, H5AC_dxpl_id) <0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_LINK, FAIL, "unable to create link");
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_API(ret_value);
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
}
/*-------------------------------------------------------------------------
* Function: H5Gunlink
*
* Purpose: Removes the specified NAME from the group graph and
* decrements the link count for the object to which NAME
* points. If the link count reaches zero then all file-space
* associated with the object will be reclaimed (but if the
* object is open, then the reclamation of the file space is
* delayed until all handles to the object are closed).
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Return: Non-negative on success/Negative on failure
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
*
* Programmer: Robb Matzke
* Monday, April 6, 1998
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
herr_t
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
H5Gunlink(hid_t loc_id, const char *name)
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
{
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
H5G_entry_t *loc = NULL;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_API(H5Gunlink, FAIL);
[svn-r429] Changes since 19980616 ---------------------- ./html/tracing.html NEW This entire update is to make it possible for the library to print the name, arguments, and return value of every API call without requiring any extra work from developers or app programmers. This file describes how this all works. ./configure.in Added the `--enable-tracing' switch. If you use it then the library will include code to print API function names, argument names and values, and function return values. However, you must then turn on the tracing by setting the HDF5_TRACE environment variable to a file descriptor number. The default is `--disable-tracing' since enabling it causes a slight increase in library size and a slowdown resulting from an extra function call for each API function call (I couldn't even measure the slowdown :-) ./bin/trace NEW A perl script that synchronizes the H5TRACE() macro calls in the *.c files with the function return type and formal argument names and types. If you use GNU make and gcc then this will be done automatically, otherwise just invoke this script with the names of one or more .c files. You could do it by hand to, but encoding argument types is a little tricky at first. ./config/commence.in Added the $(TRACE) macro, which defaults to the no-op. Added -D_POSIX_SOURCE to the compiler command line. ./src/Makefile.in Override the default for $(TRACE). ./config/depend.in Automatically calls $(TRACE) to synchronize the H5TRACE() macros in any source file that changed. As with makefile dependencies, one way to force synchronization of all files is to remove the `.depend' file. ./MANIFEST Added new files. ./src/H5Eprivate.h Modified HRETURN_ERROR() and HRETURN() for tracing. ./src/H5.c ./src/H5private.h This is where the real tracing work really happens, in H5_trace(). ./src/H5A.c ./src/H5D.c ./src/H5G.c ./src/H5P.c ./src/H5S.c ./src/H5Z.c Added H5TRACE() calls to all API functions. You don't really need these changes if you don't want to merge your stuff because they can be generated automatically by going to the hdf5/src directory and saying ../bin/trace *.c ./src/H5T.c Added H5TRACE() calls. Other stuff below. ./src/H5E.c ./src/H5Epublic.h Added H5TRACE() calls. Created a type H5E_auto_t for the `func' argument of H5Eset_auto() and H5Eget_auto() to make those arguments easier to parse for tracing. It should also make it clearer for users that don't know how to read complicated ANSI data types. ./src/H5F.c Added H5TRACE() calls. Changed a couple `uintn' argument types in API functions to `unsigned int' since `uintn' part of the API. Changed a few "can't" and "cant" error messages to "unable to". ./src/H5Ipublic.h Removed H5_DIRECTORY from the H5I_group_t enum. It wasn't used anywhere. ./src/H5Tconv.c Removed an unused label. ./src/H5Fistore.c ./src/H5Oattr.c ./src/H5Odtype.c ./src/H5T.c ./test/dsets.c ./test/dtypes.c Fixed a warning about a variable possibly used before it's initialized. Added __unused__ to turn off some unused argument warnings that pop up when debugging is turned off and optimizations are turned on.
1998-06-18 04:46:29 +08:00
H5TRACE2("e","is",loc_id,name);
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
/* Check arguments */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (NULL==(loc=H5G_loc(loc_id)))
HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, FAIL, "not a location");
if (!name || !*name)
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "no name");
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
/* Unlink */
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_unlink(loc, name, H5AC_dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, FAIL, "unable to unlink object");
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_API(ret_value);
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
}
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/*-------------------------------------------------------------------------
[svn-r654] Fixed last bug (hah! ;-) for beta release, changed H5Gget_stat to H5Gget_objinfo and renamed internal functions to match up with the renamed API functions.
1998-09-03 05:16:24 +08:00
* Function: H5Gget_objinfo
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
*
* Purpose: Returns information about an object. If FOLLOW_LINK is
* non-zero then all symbolic links are followed; otherwise all
* links except the last component of the name are followed.
*
[svn-r802] Misc. white space adjustments here and there, mostly in comments.
1998-10-27 11:56:31 +08:00
* Return: Non-negative on success, with the fields of STATBUF (if
* non-null) initialized. Negative on failure.
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
*
* Programmer: Robb Matzke
* Monday, April 13, 1998
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
herr_t
[svn-r654] Fixed last bug (hah! ;-) for beta release, changed H5Gget_stat to H5Gget_objinfo and renamed internal functions to match up with the renamed API functions.
1998-09-03 05:16:24 +08:00
H5Gget_objinfo(hid_t loc_id, const char *name, hbool_t follow_link,
[svn-r731] Changes since 19980930 ---------------------- ./src/H5G.c ./test/unlink.c Renaming a symbolic link works like on Unix: the link is renamed but the link value is unchanged.
1998-10-01 04:17:26 +08:00
H5G_stat_t *statbuf/*out*/)
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
{
[svn-r633] Changes since 19980828 ---------------------- ./RELEASE Updated with important changes I made since the second beta. ./src/H5A.c ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5R.c ./src/H5Rprivate.h ./src/H5T.c ./src/H5Tprivate.h Any API function that used to take an `hid_t loc_id' followed by a `const char *name' can now take any type of object for the loc_id as long as the object is somehow associated with a file. Internally, H5G_loc() was modified to return an H5G_entry_t* instead of an H5G_t* so it's more general. Among other things, this allows one to retrieve information about an object like a named type or dataset without knowing the name of the type or dataset: int get_nlinks (hid_t obj) { H5G_stat_t sb; if (H5Gstat(obj, ".", TRUE, &sb)<0) return -1; return sb.nlink; } ./test/gheap.c ./test/istore.c These files needed a couple of changes because they call some of the internal functions whose H5G_t arguments changed to H5G_entry_t. ./src/H5A.c Got rid of all the switch statements for getting symbol table entries for varous objects and replaced them with a call to H5G_loc() allowing attributes to automatically apply to any type of object that belongs to a file. ./test/Makefile.in Moved the ragged array tests from the normal list of tests to the `make timings' target. ./test/ragged.c Added rewrite tests -- rewrite the rows of a dataset changing the number of rows and the length of each row. ./test/mtime.c Added a test that checks that H5Gstat() can be called with a dataset as the first argument. ./src/H5S.c Added #ifdef HAVE_PARALLEL around code to check for the HDF5_MPI_OPT_TYPES environment variable because the global variable that gets set is #ifdef'd. ./bin/release bzip2 uses .bz2 as the file extension.
1998-08-31 21:46:47 +08:00
H5G_entry_t *loc = NULL;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_API(H5Gget_objinfo, FAIL);
[svn-r460] Changes since 19980707 ---------------------- ./bin/trace ./src/H5.c ./src/H5private.h ./src/H5A.c ./src/H5D.c ./src/H5F.c ./src/H5G.c ./src/H5P.c ./src/H5Pprivate.h ./src/H5S.c ./src/H5T.c Output-only arguments have their addresses printed during tracing and added symbolic output for the H5F_driver_t arguments. That's another reason that we should be careful to add `/*out*/' after arguments that are output-only and `/*in,out*/' after arguments that are used for both input and output values. No internal function calls H5Pget_class() anymore. ./src/H5T.c ./src/H5Tconv.c ./src/H5Tpkg.h ./src/H5Tpublic.h Added H5Tget_overflow() and H5Tset_overflow() so the application can query or set a function that will be called whenever an overflow occurs. Implemented as documented in previous e-mail except the overflow handler gets two buffers: one that contains the source value and one to receive the optional destination value. ./test/dtypes.c Tests overflow handler. ./src/H5.c We have to declare fdopen() because I'm getting errors when compiling on Irix64 even though we include <stdio.h> as documented in the fdopen() man page.
1998-07-08 04:13:31 +08:00
H5TRACE4("e","isbx",loc_id,name,follow_link,statbuf);
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/* Check arguments */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (NULL==(loc=H5G_loc (loc_id)))
HGOTO_ERROR (H5E_ARGS, H5E_BADTYPE, FAIL, "not a location");
if (!name || !*name)
HGOTO_ERROR (H5E_ARGS, H5E_BADVALUE, FAIL, "no name specified");
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/* Get info */
[svn-r6388] Purpose: Update feature Description: Relax collective constraint for API functions which only read metadata from a file. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 w/parallel
2003-02-11 02:44:22 +08:00
if (H5G_get_objinfo (loc, name, follow_link, statbuf, H5AC_ind_dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_ARGS, H5E_CANTINIT, FAIL, "cannot stat object");
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_API(ret_value);
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
}
/*-------------------------------------------------------------------------
* Function: H5Gget_linkval
*
* Purpose: Returns the value of a symbolic link whose name is NAME. At
* most SIZE characters (counting the null terminator) are
* copied to the BUF result buffer.
*
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
* Return: Success: Non-negative with the link value in BUF.
*
* Failure: Negative
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
*
* Programmer: Robb Matzke
* Monday, April 13, 1998
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
herr_t
[svn-r620] Changes since 19980825 ---------------------- ./MANIFEST ./src/H5R.c [NEW] ./src/H5Rprivate.h [NEW] ./src/H5Rpublic.h [NEW] ./src/Makefile.in ./src/hdf5.h ./test/ragged.c [NEW] Preliminary support for 2d ragged arrays for Mark Miller and Jim Reus. Not fully implemented yet. The test is not actually part of `make test' because we still have some memory problems. ./src/H5E.c ./src/H5Epublic.h Added H5E_RAGGED as a major error number. ./bin/release Checks the MANIFEST file against `svf ls' on systems that have it. ./bin/trace Fixed a bug that caused arguments of type `void *x[]' to not be handled. ./src/H5.c Removed unused variables and changed a couple types to fix compiler warnings. Added tracing support for ragged array object ID's and arrays of pointers. ./src/H5D.c H5Dcreate() will complain if either of the property lists are invalid (instead of using the default). ./src/H5D.c ./src/H5Dprivate.h Split H5Dget_space() into an API and internal function so it can be called from the new ragged array layer. ./src/H5Fistore.c Fixed warnings about unsigned vs. signed comparisons. ./src/H5Flow.c Fixed a warning about a variable being shadowed in the MPI-IO stuff. ./src/H5Iprivate.h ./src/H5Ipublic.h Added the H5_RAGGED atom group. ./src/H5Shyper.c Fixed some freeing-free-memory errors that resulted when certain arrays were freed but the pointers were left in the data structures. I simply set the pointers to null after they were freed. ./src/H5Sprivate.h ./src/H5Sselect.c Split the H5Sselect_hyperslab() function into an API and a private function so it could be called from the ragged array layer. Added H5S_SEL_ERROR and H5S_SEL_N to the switch statements to get rid or compiler warnings. ./src/H5Tconv.c Removed a misleading comment. ./test/bittests.c Fixed a warning about a printf(). ./test/cmpd_dset.c Fixed warnings about unused variables because of test #11 being commented out. ./bin/trace Shortened the right margin for the output to allow room for the `);' at the end of the TRACE() macros.
1998-08-28 00:48:50 +08:00
H5Gget_linkval(hid_t loc_id, const char *name, size_t size, char *buf/*out*/)
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
{
[svn-r633] Changes since 19980828 ---------------------- ./RELEASE Updated with important changes I made since the second beta. ./src/H5A.c ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5R.c ./src/H5Rprivate.h ./src/H5T.c ./src/H5Tprivate.h Any API function that used to take an `hid_t loc_id' followed by a `const char *name' can now take any type of object for the loc_id as long as the object is somehow associated with a file. Internally, H5G_loc() was modified to return an H5G_entry_t* instead of an H5G_t* so it's more general. Among other things, this allows one to retrieve information about an object like a named type or dataset without knowing the name of the type or dataset: int get_nlinks (hid_t obj) { H5G_stat_t sb; if (H5Gstat(obj, ".", TRUE, &sb)<0) return -1; return sb.nlink; } ./test/gheap.c ./test/istore.c These files needed a couple of changes because they call some of the internal functions whose H5G_t arguments changed to H5G_entry_t. ./src/H5A.c Got rid of all the switch statements for getting symbol table entries for varous objects and replaced them with a call to H5G_loc() allowing attributes to automatically apply to any type of object that belongs to a file. ./test/Makefile.in Moved the ragged array tests from the normal list of tests to the `make timings' target. ./test/ragged.c Added rewrite tests -- rewrite the rows of a dataset changing the number of rows and the length of each row. ./test/mtime.c Added a test that checks that H5Gstat() can be called with a dataset as the first argument. ./src/H5S.c Added #ifdef HAVE_PARALLEL around code to check for the HDF5_MPI_OPT_TYPES environment variable because the global variable that gets set is #ifdef'd. ./bin/release bzip2 uses .bz2 as the file extension.
1998-08-31 21:46:47 +08:00
H5G_entry_t *loc = NULL;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_API(H5Gget_linkval, FAIL);
[svn-r460] Changes since 19980707 ---------------------- ./bin/trace ./src/H5.c ./src/H5private.h ./src/H5A.c ./src/H5D.c ./src/H5F.c ./src/H5G.c ./src/H5P.c ./src/H5Pprivate.h ./src/H5S.c ./src/H5T.c Output-only arguments have their addresses printed during tracing and added symbolic output for the H5F_driver_t arguments. That's another reason that we should be careful to add `/*out*/' after arguments that are output-only and `/*in,out*/' after arguments that are used for both input and output values. No internal function calls H5Pget_class() anymore. ./src/H5T.c ./src/H5Tconv.c ./src/H5Tpkg.h ./src/H5Tpublic.h Added H5Tget_overflow() and H5Tset_overflow() so the application can query or set a function that will be called whenever an overflow occurs. Implemented as documented in previous e-mail except the overflow handler gets two buffers: one that contains the source value and one to receive the optional destination value. ./test/dtypes.c Tests overflow handler. ./src/H5.c We have to declare fdopen() because I'm getting errors when compiling on Irix64 even though we include <stdio.h> as documented in the fdopen() man page.
1998-07-08 04:13:31 +08:00
H5TRACE4("e","iszx",loc_id,name,size,buf);
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/* Check arguments */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (NULL==(loc=H5G_loc (loc_id)))
HGOTO_ERROR (H5E_ARGS, H5E_BADTYPE, FAIL, "not a location");
if (!name || !*name)
HGOTO_ERROR (H5E_ARGS, H5E_BADVALUE, FAIL, "no name specified");
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/* Get the link value */
[svn-r6388] Purpose: Update feature Description: Relax collective constraint for API functions which only read metadata from a file. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 w/parallel
2003-02-11 02:44:22 +08:00
if (H5G_linkval (loc, name, size, buf, H5AC_ind_dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_NOTFOUND, FAIL, "unable to get link value");
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_API(ret_value);
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
}
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
/*-------------------------------------------------------------------------
* Function: H5Gset_comment
*
* Purpose: Gives the specified object a comment. The COMMENT string
* should be a null terminated string. An object can have only
* one comment at a time. Passing NULL for the COMMENT argument
* will remove the comment property from the object.
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Return: Non-negative on success/Negative on failure
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
*
* Programmer: Robb Matzke
* Monday, July 20, 1998
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
herr_t
[svn-r620] Changes since 19980825 ---------------------- ./MANIFEST ./src/H5R.c [NEW] ./src/H5Rprivate.h [NEW] ./src/H5Rpublic.h [NEW] ./src/Makefile.in ./src/hdf5.h ./test/ragged.c [NEW] Preliminary support for 2d ragged arrays for Mark Miller and Jim Reus. Not fully implemented yet. The test is not actually part of `make test' because we still have some memory problems. ./src/H5E.c ./src/H5Epublic.h Added H5E_RAGGED as a major error number. ./bin/release Checks the MANIFEST file against `svf ls' on systems that have it. ./bin/trace Fixed a bug that caused arguments of type `void *x[]' to not be handled. ./src/H5.c Removed unused variables and changed a couple types to fix compiler warnings. Added tracing support for ragged array object ID's and arrays of pointers. ./src/H5D.c H5Dcreate() will complain if either of the property lists are invalid (instead of using the default). ./src/H5D.c ./src/H5Dprivate.h Split H5Dget_space() into an API and internal function so it can be called from the new ragged array layer. ./src/H5Fistore.c Fixed warnings about unsigned vs. signed comparisons. ./src/H5Flow.c Fixed a warning about a variable being shadowed in the MPI-IO stuff. ./src/H5Iprivate.h ./src/H5Ipublic.h Added the H5_RAGGED atom group. ./src/H5Shyper.c Fixed some freeing-free-memory errors that resulted when certain arrays were freed but the pointers were left in the data structures. I simply set the pointers to null after they were freed. ./src/H5Sprivate.h ./src/H5Sselect.c Split the H5Sselect_hyperslab() function into an API and a private function so it could be called from the ragged array layer. Added H5S_SEL_ERROR and H5S_SEL_N to the switch statements to get rid or compiler warnings. ./src/H5Tconv.c Removed a misleading comment. ./test/bittests.c Fixed a warning about a printf(). ./test/cmpd_dset.c Fixed warnings about unused variables because of test #11 being commented out. ./bin/trace Shortened the right margin for the output to allow room for the `);' at the end of the TRACE() macros.
1998-08-28 00:48:50 +08:00
H5Gset_comment(hid_t loc_id, const char *name, const char *comment)
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
{
[svn-r633] Changes since 19980828 ---------------------- ./RELEASE Updated with important changes I made since the second beta. ./src/H5A.c ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5R.c ./src/H5Rprivate.h ./src/H5T.c ./src/H5Tprivate.h Any API function that used to take an `hid_t loc_id' followed by a `const char *name' can now take any type of object for the loc_id as long as the object is somehow associated with a file. Internally, H5G_loc() was modified to return an H5G_entry_t* instead of an H5G_t* so it's more general. Among other things, this allows one to retrieve information about an object like a named type or dataset without knowing the name of the type or dataset: int get_nlinks (hid_t obj) { H5G_stat_t sb; if (H5Gstat(obj, ".", TRUE, &sb)<0) return -1; return sb.nlink; } ./test/gheap.c ./test/istore.c These files needed a couple of changes because they call some of the internal functions whose H5G_t arguments changed to H5G_entry_t. ./src/H5A.c Got rid of all the switch statements for getting symbol table entries for varous objects and replaced them with a call to H5G_loc() allowing attributes to automatically apply to any type of object that belongs to a file. ./test/Makefile.in Moved the ragged array tests from the normal list of tests to the `make timings' target. ./test/ragged.c Added rewrite tests -- rewrite the rows of a dataset changing the number of rows and the length of each row. ./test/mtime.c Added a test that checks that H5Gstat() can be called with a dataset as the first argument. ./src/H5S.c Added #ifdef HAVE_PARALLEL around code to check for the HDF5_MPI_OPT_TYPES environment variable because the global variable that gets set is #ifdef'd. ./bin/release bzip2 uses .bz2 as the file extension.
1998-08-31 21:46:47 +08:00
H5G_entry_t *loc = NULL;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_API(H5Gset_comment, FAIL);
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
H5TRACE3("e","iss",loc_id,name,comment);
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (NULL==(loc=H5G_loc(loc_id)))
HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, FAIL, "not a location");
if (!name || !*name)
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "no name specified");
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_set_comment(loc, name, comment, H5AC_dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, FAIL, "unable to set comment value");
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_API(ret_value);
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
}
/*-------------------------------------------------------------------------
* Function: H5Gget_comment
*
* Purpose: Return at most BUFSIZE characters of the comment for the
* specified object. If BUFSIZE is large enough to hold the
* entire comment then the comment string will be null
* terminated, otherwise it will not. If the object does not
[svn-r537] Changes since 19980722 ---------------------- ./src/H5A.c ./src/H5Apublic.h ./test/tattr.c Switched the order of the second and third argument of H5Aget_name() to make it consistent with other functions that take buffers and buffer sizes. ./src/H5G.c ./src/H5Gpublic.h ./src/H5Gprivate.h The H5Gget_comment() function returns the size of the comment including the null terminator. If the object has no comment then zero is returned. If an error occurs then a negative value is returned. ./MANIFEST ./tools/Makefile.in ./tools/h5tools.h [NEW] ./tools/h5dump.c [NEW] Created a library for printing values of datasets in a way that looks nice. It's not done yet, but I needed it for debugging the contents of files from Jim Reus. ./tools/h5ls.c Added the `-d' and `--dump' options which cause the contents of a dataset to be printed. Added `-w N' and `--width=N' options to control how wide the raw data output should be. If you want single-column output then say `-w1'. Printing dataset values can now handle datasets of any integer or floating point atomic type. As a special case, integers which are one byte wide are treated a character strings for now. Sample output: $ h5ls --dump --width=60 banana.hdf ARCHIVE 0:0:0:744 Dataset {52/Inf} Data: (0) "U struct complex { double R; double I; };\012V" (43) " double;\012" U 0:0:0:2500 Dataset {256/512} Data: printing of compound data types is not implemented yet V 0:0:0:3928 Dataset {256/512} Data: (0) 0, 0.015625, 0.03125, 0.046875, 0.0625, (5) 0.078125, 0.09375, 0.109375, 0.125, 0.140625, (10) 0.15625, 0.171875, 0.1875, 0.203125, 0.21875, (15) 0.234375, 0.25, 0.265625, 0.28125, 0.296875, ...
1998-07-24 05:19:17 +08:00
* have a comment value then no bytes are copied to the BUF
* buffer.
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
*
[svn-r537] Changes since 19980722 ---------------------- ./src/H5A.c ./src/H5Apublic.h ./test/tattr.c Switched the order of the second and third argument of H5Aget_name() to make it consistent with other functions that take buffers and buffer sizes. ./src/H5G.c ./src/H5Gpublic.h ./src/H5Gprivate.h The H5Gget_comment() function returns the size of the comment including the null terminator. If the object has no comment then zero is returned. If an error occurs then a negative value is returned. ./MANIFEST ./tools/Makefile.in ./tools/h5tools.h [NEW] ./tools/h5dump.c [NEW] Created a library for printing values of datasets in a way that looks nice. It's not done yet, but I needed it for debugging the contents of files from Jim Reus. ./tools/h5ls.c Added the `-d' and `--dump' options which cause the contents of a dataset to be printed. Added `-w N' and `--width=N' options to control how wide the raw data output should be. If you want single-column output then say `-w1'. Printing dataset values can now handle datasets of any integer or floating point atomic type. As a special case, integers which are one byte wide are treated a character strings for now. Sample output: $ h5ls --dump --width=60 banana.hdf ARCHIVE 0:0:0:744 Dataset {52/Inf} Data: (0) "U struct complex { double R; double I; };\012V" (43) " double;\012" U 0:0:0:2500 Dataset {256/512} Data: printing of compound data types is not implemented yet V 0:0:0:3928 Dataset {256/512} Data: (0) 0, 0.015625, 0.03125, 0.046875, 0.0625, (5) 0.078125, 0.09375, 0.109375, 0.125, 0.140625, (10) 0.15625, 0.171875, 0.1875, 0.203125, 0.21875, (15) 0.234375, 0.25, 0.265625, 0.28125, 0.296875, ...
1998-07-24 05:19:17 +08:00
* Return: Success: Number of characters in the comment counting
* the null terminator. The value returned may
* be larger than the BUFSIZE argument.
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Failure: Negative
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
*
* Programmer: Robb Matzke
* Monday, July 20, 1998
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
[svn-r537] Changes since 19980722 ---------------------- ./src/H5A.c ./src/H5Apublic.h ./test/tattr.c Switched the order of the second and third argument of H5Aget_name() to make it consistent with other functions that take buffers and buffer sizes. ./src/H5G.c ./src/H5Gpublic.h ./src/H5Gprivate.h The H5Gget_comment() function returns the size of the comment including the null terminator. If the object has no comment then zero is returned. If an error occurs then a negative value is returned. ./MANIFEST ./tools/Makefile.in ./tools/h5tools.h [NEW] ./tools/h5dump.c [NEW] Created a library for printing values of datasets in a way that looks nice. It's not done yet, but I needed it for debugging the contents of files from Jim Reus. ./tools/h5ls.c Added the `-d' and `--dump' options which cause the contents of a dataset to be printed. Added `-w N' and `--width=N' options to control how wide the raw data output should be. If you want single-column output then say `-w1'. Printing dataset values can now handle datasets of any integer or floating point atomic type. As a special case, integers which are one byte wide are treated a character strings for now. Sample output: $ h5ls --dump --width=60 banana.hdf ARCHIVE 0:0:0:744 Dataset {52/Inf} Data: (0) "U struct complex { double R; double I; };\012V" (43) " double;\012" U 0:0:0:2500 Dataset {256/512} Data: printing of compound data types is not implemented yet V 0:0:0:3928 Dataset {256/512} Data: (0) 0, 0.015625, 0.03125, 0.046875, 0.0625, (5) 0.078125, 0.09375, 0.109375, 0.125, 0.140625, (10) 0.15625, 0.171875, 0.1875, 0.203125, 0.21875, (15) 0.234375, 0.25, 0.265625, 0.28125, 0.296875, ...
1998-07-24 05:19:17 +08:00
int
[svn-r620] Changes since 19980825 ---------------------- ./MANIFEST ./src/H5R.c [NEW] ./src/H5Rprivate.h [NEW] ./src/H5Rpublic.h [NEW] ./src/Makefile.in ./src/hdf5.h ./test/ragged.c [NEW] Preliminary support for 2d ragged arrays for Mark Miller and Jim Reus. Not fully implemented yet. The test is not actually part of `make test' because we still have some memory problems. ./src/H5E.c ./src/H5Epublic.h Added H5E_RAGGED as a major error number. ./bin/release Checks the MANIFEST file against `svf ls' on systems that have it. ./bin/trace Fixed a bug that caused arguments of type `void *x[]' to not be handled. ./src/H5.c Removed unused variables and changed a couple types to fix compiler warnings. Added tracing support for ragged array object ID's and arrays of pointers. ./src/H5D.c H5Dcreate() will complain if either of the property lists are invalid (instead of using the default). ./src/H5D.c ./src/H5Dprivate.h Split H5Dget_space() into an API and internal function so it can be called from the new ragged array layer. ./src/H5Fistore.c Fixed warnings about unsigned vs. signed comparisons. ./src/H5Flow.c Fixed a warning about a variable being shadowed in the MPI-IO stuff. ./src/H5Iprivate.h ./src/H5Ipublic.h Added the H5_RAGGED atom group. ./src/H5Shyper.c Fixed some freeing-free-memory errors that resulted when certain arrays were freed but the pointers were left in the data structures. I simply set the pointers to null after they were freed. ./src/H5Sprivate.h ./src/H5Sselect.c Split the H5Sselect_hyperslab() function into an API and a private function so it could be called from the ragged array layer. Added H5S_SEL_ERROR and H5S_SEL_N to the switch statements to get rid or compiler warnings. ./src/H5Tconv.c Removed a misleading comment. ./test/bittests.c Fixed a warning about a printf(). ./test/cmpd_dset.c Fixed warnings about unused variables because of test #11 being commented out. ./bin/trace Shortened the right margin for the output to allow room for the `);' at the end of the TRACE() macros.
1998-08-28 00:48:50 +08:00
H5Gget_comment(hid_t loc_id, const char *name, size_t bufsize, char *buf)
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
{
[svn-r633] Changes since 19980828 ---------------------- ./RELEASE Updated with important changes I made since the second beta. ./src/H5A.c ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5R.c ./src/H5Rprivate.h ./src/H5T.c ./src/H5Tprivate.h Any API function that used to take an `hid_t loc_id' followed by a `const char *name' can now take any type of object for the loc_id as long as the object is somehow associated with a file. Internally, H5G_loc() was modified to return an H5G_entry_t* instead of an H5G_t* so it's more general. Among other things, this allows one to retrieve information about an object like a named type or dataset without knowing the name of the type or dataset: int get_nlinks (hid_t obj) { H5G_stat_t sb; if (H5Gstat(obj, ".", TRUE, &sb)<0) return -1; return sb.nlink; } ./test/gheap.c ./test/istore.c These files needed a couple of changes because they call some of the internal functions whose H5G_t arguments changed to H5G_entry_t. ./src/H5A.c Got rid of all the switch statements for getting symbol table entries for varous objects and replaced them with a call to H5G_loc() allowing attributes to automatically apply to any type of object that belongs to a file. ./test/Makefile.in Moved the ragged array tests from the normal list of tests to the `make timings' target. ./test/ragged.c Added rewrite tests -- rewrite the rows of a dataset changing the number of rows and the length of each row. ./test/mtime.c Added a test that checks that H5Gstat() can be called with a dataset as the first argument. ./src/H5S.c Added #ifdef HAVE_PARALLEL around code to check for the HDF5_MPI_OPT_TYPES environment variable because the global variable that gets set is #ifdef'd. ./bin/release bzip2 uses .bz2 as the file extension.
1998-08-31 21:46:47 +08:00
H5G_entry_t *loc = NULL;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
int ret_value;
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_API(H5Gget_comment, FAIL);
[svn-r537] Changes since 19980722 ---------------------- ./src/H5A.c ./src/H5Apublic.h ./test/tattr.c Switched the order of the second and third argument of H5Aget_name() to make it consistent with other functions that take buffers and buffer sizes. ./src/H5G.c ./src/H5Gpublic.h ./src/H5Gprivate.h The H5Gget_comment() function returns the size of the comment including the null terminator. If the object has no comment then zero is returned. If an error occurs then a negative value is returned. ./MANIFEST ./tools/Makefile.in ./tools/h5tools.h [NEW] ./tools/h5dump.c [NEW] Created a library for printing values of datasets in a way that looks nice. It's not done yet, but I needed it for debugging the contents of files from Jim Reus. ./tools/h5ls.c Added the `-d' and `--dump' options which cause the contents of a dataset to be printed. Added `-w N' and `--width=N' options to control how wide the raw data output should be. If you want single-column output then say `-w1'. Printing dataset values can now handle datasets of any integer or floating point atomic type. As a special case, integers which are one byte wide are treated a character strings for now. Sample output: $ h5ls --dump --width=60 banana.hdf ARCHIVE 0:0:0:744 Dataset {52/Inf} Data: (0) "U struct complex { double R; double I; };\012V" (43) " double;\012" U 0:0:0:2500 Dataset {256/512} Data: printing of compound data types is not implemented yet V 0:0:0:3928 Dataset {256/512} Data: (0) 0, 0.015625, 0.03125, 0.046875, 0.0625, (5) 0.078125, 0.09375, 0.109375, 0.125, 0.140625, (10) 0.15625, 0.171875, 0.1875, 0.203125, 0.21875, (15) 0.234375, 0.25, 0.265625, 0.28125, 0.296875, ...
1998-07-24 05:19:17 +08:00
H5TRACE4("Is","iszs",loc_id,name,bufsize,buf);
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (NULL==(loc=H5G_loc(loc_id)))
HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, FAIL, "not a location");
if (!name || !*name)
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "no name specified");
if (bufsize>0 && !buf)
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "no buffer specified");
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
[svn-r6388] Purpose: Update feature Description: Relax collective constraint for API functions which only read metadata from a file. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 w/parallel
2003-02-11 02:44:22 +08:00
if ((ret_value=H5G_get_comment(loc, name, bufsize, buf, H5AC_ind_dxpl_id))<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, FAIL, "unable to get comment value");
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_API(ret_value);
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
}
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
/*
*-------------------------------------------------------------------------
*-------------------------------------------------------------------------
* N O A P I F U N C T I O N S B E Y O N D T H I S P O I N T
*-------------------------------------------------------------------------
*-------------------------------------------------------------------------
*/
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
/*-------------------------------------------------------------------------
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Function: H5G_init_interface
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Purpose: Initializes the H5G interface.
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Return: Non-negative on success/Negative on failure
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Programmer: Robb Matzke
* Monday, January 5, 1998
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
static herr_t
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
H5G_init_interface(void)
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
{
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r5467] Purpose: Code cleanup. Description: Took Robb's recent ideas for improving the FUNC_ENTER/FUNC_LEAVE macros equivalents in the SAF library and adapted them to our library. I added an additional macro which is equivalent to FUNC_ENTER: FUNC_ENTER_NOINIT - Has the API tracing code, etc. from FUNC_ENTER but none of the library or interface initialization code. This is to be used _only_ for static functions and those which explicitly cannot have the library or interface initialization code enabled (like the API termination routines, etc.). This allowed many more of the functions in the library [but not all yet :-(] to be wrapped with FUNC_ENTER[_NOINIT]/FUNC_LEAVE pairs. It also reduced the size of the library and executables (by cutting out a bunch of code which was never executed), I'll e-mail the exact results when I've finished editing it. Platforms tested: IRIX64 6.5 (modi4)
2002-05-29 02:17:12 +08:00
FUNC_ENTER_NOINIT(H5G_init_interface);
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/* Initialize the atom group for the group IDs */
[svn-r737] Changed H5I groups to have a H5I_ prefix before them, to better align with the library.
1998-10-06 07:41:30 +08:00
if (H5I_init_group(H5I_GROUP, H5I_GROUPID_HASHSIZE, H5G_RESERVED_ATOMS,
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
(H5I_free_t)H5G_close) < 0)
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, FAIL, "unable to initialize interface");
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
/*
* Initialize the type info table. Begin with the most general types and
* end with the most specific. For instance, any object that has a data
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
* type message is a datatype but only some of them are datasets.
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
*/
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
H5G_register_type(H5G_TYPE, H5T_isa, "datatype");
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
H5G_register_type(H5G_GROUP, H5G_isa, "group");
H5G_register_type(H5G_DATASET, H5D_isa, "dataset");
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
}
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
/*-------------------------------------------------------------------------
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Function: H5G_term_interface
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Purpose: Terminates the H5G interface
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r1169] ./configure.in ./configure [REGENERATED] ./src/H5D.c ./src/H5O.c Removed H5AC, H5B, and H5T from the default list of packages to debug (because they're pretty expensive debugging), and added H5O. Also fixed a bug for undefined variable in H5D when H5S debugging is turned on but H5T debugging is turned off. ./config/conclude.in Fixed installation of header files for building in a directory other than the source directory. This fixes a bug where H5config.h wasn't being installed. ./src/H5.c ./src/H5A.c ./src/H5D.c ./src/H5F.c ./src/H5G.c ./src/H5I.c ./src/H5Iprivate.h ./src/H5P.c ./src/H5R.c ./src/H5RA.c ./src/H5S.c ./src/H5T.c ./src/H5TB.c ./src/H5Tprivate.h ./src/H5Z.c ./src/H5detect.c ./src/H5private.h Changed the way the library shuts down again. Now it handles cycles between packages and isn't so sensitive to dependencies between packages. A package might shut down only to be restarted to process a request from some other package being shut down. Loops are detected after 100 iteractions and the shutdown is aborted with a message on standard error. This also makes it a lot easier to debug. ./src/H5A.c Fixed H5A_write() and H5A_read() so they pass a non-null background buffer to the conversion functions. This is necessary when an attribute has a compound data type. ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fsec2.c Reindented new Win32 stuff. ./src/H5Odtype.c Fixed a bug when enumeration types are used in a compound data type. The byte pointer wasn't incremented after the type information was written. ./tools/h5ls.c Compound data types display their total size because it's not always obvious from looking at the members. Scalar attributes show their space as `scalar' instead of `{}'. The index value is not printed for attributes that have only a few values. Instead the word `Data:' is printed on the first line of attribute data. Named types display their data type only if verbose output was requested.
1999-03-30 19:38:34 +08:00
* Return: Success: Positive if anything is done that might
* affect other interfaces; zero otherwise.
*
* Failure: Negative.
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Programmer: Robb Matzke
* Monday, January 5, 1998
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
* Modifications:
[svn-r5110] ./hdf5-devel/src/H5G.c Merge of name component length limit bugfix from 1.4 branch. A couple of very minor typos fixed.
2002-03-29 03:27:38 +08:00
* Robb Matzke, 2002-03-28
* Free the global component buffer.
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*-------------------------------------------------------------------------
*/
[svn-r4355] Purpose: Code cleanup (sorta) Description: When the first versions of the HDF5 library were designed, I remembered vividly the difficulties of porting code from a 32-bit platform to a 16-bit platform and asked that people use intn & uintn instead of int & unsigned int, respectively. However, in hindsight, this was overkill and unnecessary since we weren't going to be porting the HDF5 library to 16-bit architectures. Currently, the extra uintn & intn typedefs are causing problems for users who'd like to include both the HDF5 and HDF4 header files in one source module (like Kent's h4toh5 library). Solution: Changed the uintn & intn's to unsigned and int's respectively. Platforms tested: FreeBSD 4.4 (hawkwind)
2001-08-15 06:09:56 +08:00
int
[svn-r1169] ./configure.in ./configure [REGENERATED] ./src/H5D.c ./src/H5O.c Removed H5AC, H5B, and H5T from the default list of packages to debug (because they're pretty expensive debugging), and added H5O. Also fixed a bug for undefined variable in H5D when H5S debugging is turned on but H5T debugging is turned off. ./config/conclude.in Fixed installation of header files for building in a directory other than the source directory. This fixes a bug where H5config.h wasn't being installed. ./src/H5.c ./src/H5A.c ./src/H5D.c ./src/H5F.c ./src/H5G.c ./src/H5I.c ./src/H5Iprivate.h ./src/H5P.c ./src/H5R.c ./src/H5RA.c ./src/H5S.c ./src/H5T.c ./src/H5TB.c ./src/H5Tprivate.h ./src/H5Z.c ./src/H5detect.c ./src/H5private.h Changed the way the library shuts down again. Now it handles cycles between packages and isn't so sensitive to dependencies between packages. A package might shut down only to be restarted to process a request from some other package being shut down. Loops are detected after 100 iteractions and the shutdown is aborted with a message on standard error. This also makes it a lot easier to debug. ./src/H5A.c Fixed H5A_write() and H5A_read() so they pass a non-null background buffer to the conversion functions. This is necessary when an attribute has a compound data type. ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fsec2.c Reindented new Win32 stuff. ./src/H5Odtype.c Fixed a bug when enumeration types are used in a compound data type. The byte pointer wasn't incremented after the type information was written. ./tools/h5ls.c Compound data types display their total size because it's not always obvious from looking at the members. Scalar attributes show their space as `scalar' instead of `{}'. The index value is not printed for attributes that have only a few values. Instead the word `Data:' is printed on the first line of attribute data. Named types display their data type only if verbose output was requested.
1999-03-30 19:38:34 +08:00
H5G_term_interface(void)
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
{
[svn-r914] Changes since 19981113 ---------------------- ./config/conclude.in ./test/Makefile.in ./tools/Makefile.in The Makefile $TESTS variable has finally been split into $TEST_PROGS and $TEST_SCRIPTS with the latter being the names of shell scripts that need to be run with `/bin/sh'. Now we don't have to copy each shell script before we run it. NOTE: THIS CHANGE DOES NOT AFFECT THE PABLO MAKEFILE SINCE THAT FILE IS A COPY OF A PREVIOUSLY GENERATED MAKEFILE. ./src/Makefile.in Added H5Snone.c to the source list. ./src/H5G.c Plugged a memory leak by emptying the object type "isa" table when the library is closed. ./src/H5Tconv.c ./src/H5Tpkg.h Added 48 new type functions to take advantage of hardware for integer type conversions. These functions are not registered in the conversion table yet because I haven't fully tested them. ./src/H5Tpkg.h Removed __unused__ qualifiers from prototypes.
1998-11-16 23:29:54 +08:00
size_t i;
[svn-r4355] Purpose: Code cleanup (sorta) Description: When the first versions of the HDF5 library were designed, I remembered vividly the difficulties of porting code from a 32-bit platform to a 16-bit platform and asked that people use intn & uintn instead of int & unsigned int, respectively. However, in hindsight, this was overkill and unnecessary since we weren't going to be porting the HDF5 library to 16-bit architectures. Currently, the extra uintn & intn typedefs are causing problems for users who'd like to include both the HDF5 and HDF4 header files in one source module (like Kent's h4toh5 library). Solution: Changed the uintn & intn's to unsigned and int's respectively. Platforms tested: FreeBSD 4.4 (hawkwind)
2001-08-15 06:09:56 +08:00
int n=0;
[svn-r5467] Purpose: Code cleanup. Description: Took Robb's recent ideas for improving the FUNC_ENTER/FUNC_LEAVE macros equivalents in the SAF library and adapted them to our library. I added an additional macro which is equivalent to FUNC_ENTER: FUNC_ENTER_NOINIT - Has the API tracing code, etc. from FUNC_ENTER but none of the library or interface initialization code. This is to be used _only_ for static functions and those which explicitly cannot have the library or interface initialization code enabled (like the API termination routines, etc.). This allowed many more of the functions in the library [but not all yet :-(] to be wrapped with FUNC_ENTER[_NOINIT]/FUNC_LEAVE pairs. It also reduced the size of the library and executables (by cutting out a bunch of code which was never executed), I'll e-mail the exact results when I've finished editing it. Platforms tested: IRIX64 6.5 (modi4)
2002-05-29 02:17:12 +08:00
FUNC_ENTER_NOINIT(H5G_term_interface);
[svn-r1169] ./configure.in ./configure [REGENERATED] ./src/H5D.c ./src/H5O.c Removed H5AC, H5B, and H5T from the default list of packages to debug (because they're pretty expensive debugging), and added H5O. Also fixed a bug for undefined variable in H5D when H5S debugging is turned on but H5T debugging is turned off. ./config/conclude.in Fixed installation of header files for building in a directory other than the source directory. This fixes a bug where H5config.h wasn't being installed. ./src/H5.c ./src/H5A.c ./src/H5D.c ./src/H5F.c ./src/H5G.c ./src/H5I.c ./src/H5Iprivate.h ./src/H5P.c ./src/H5R.c ./src/H5RA.c ./src/H5S.c ./src/H5T.c ./src/H5TB.c ./src/H5Tprivate.h ./src/H5Z.c ./src/H5detect.c ./src/H5private.h Changed the way the library shuts down again. Now it handles cycles between packages and isn't so sensitive to dependencies between packages. A package might shut down only to be restarted to process a request from some other package being shut down. Loops are detected after 100 iteractions and the shutdown is aborted with a message on standard error. This also makes it a lot easier to debug. ./src/H5A.c Fixed H5A_write() and H5A_read() so they pass a non-null background buffer to the conversion functions. This is necessary when an attribute has a compound data type. ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fsec2.c Reindented new Win32 stuff. ./src/H5Odtype.c Fixed a bug when enumeration types are used in a compound data type. The byte pointer wasn't incremented after the type information was written. ./tools/h5ls.c Compound data types display their total size because it's not always obvious from looking at the members. Scalar attributes show their space as `scalar' instead of `{}'. The index value is not printed for attributes that have only a few values. Instead the word `Data:' is printed on the first line of attribute data. Named types display their data type only if verbose output was requested.
1999-03-30 19:38:34 +08:00
if (interface_initialize_g) {
if ((n=H5I_nmembers(H5I_GROUP))) {
[svn-r1224] Changes since 19990426 ---------------------- ./tools/h5tools.c ./tools/h5tools.h Finally fixed a long-standing bug that caused core dumps if a compound datum rendered to more than some number of characters (we kept bumping up the limit at the risk of violating stack size limits on some machines). The fix works only on systems that have the vsnprintf() function (otherwise a 4kB limit is imposed, which if violated probably dumps core). If vsnprintf() is present then the library dynamically allocates space for the output string. Also made it possible to control how compound data is rendered across multiple lines of output by allowing the caller to specify where optional line-breaks get inserted. The output functions split up the value at one or more optional line-breaks to prevent it from wrapping around the screen. If a datum doesn't fit on the current line but would fit on the next line then it is printed on the next line regardless of whether optional line-breaks would have prevent wrapping around the screen. This makes it easier to find the beginnings of compound data values. This feature is disabled by default but can be enabled by the application. If a datum doesn't fit on the current line and the previous datum also occupied more than one line then we move to the next line before printing. This makes it easier to find the beginnings of compound data values but prevents the output from looking fragmented if there are only a few long values among mostly short values. This feature is disabled by default but can be enabled by the application. The application can control the printf() formats used for all the native data types. The defaults are what the library used to use: %g, %ld, %lu, %d, and %u ./tools/h5ls.c Compound datatype values can now be split across multiple lines of output instead of just wrapping. Also, when lots of compound values are too long they all start at the beginning of a line. This only required about 10 lines of changes in the setup for tools library calls (I didn't modify the h5dump program because it uses its own version of the tools library that forked off long ago). Added code for Win32 which is unable to cast `unsigned long long' to `double'. If the dataset size exceeds (2^63)-1 then the percent utilization is not displayed (this is easily possible with chunked datasets). This is untested yet. ./configure.in ./src/H5config.h.in ./src/H5.c ./src/H5private.h Check for vsnprintf() and provide a simple, stupid definition if it isn't available. The stupid definition just calls vsprintf() and ignores the second argument. This can result in buffer overflows in h5ls and h5dump since vsprintf() is an unsafe function (and anyone can create an hdf5 file that runs an arbitrary command from h5ls and h5dump in that case)! ./config/conclude.in Remove more *.o files for `make clean' ./src/H5A.c ./src/H5D.c ./src/H5F.c ./src/H5I.c ./src/H5Iprivate.h ./src/H5P.c ./src/H5R.c ./src/H5RA.c ./src/H5S.c ./src/H5T.c ./src/H5TB.c Cleaned up a memory leak during H5_term_library() by allowing H5I_clear_group() to skip items that couldn't be freed. This allows the item to remain in the group until we can free it later. ./src/H5F.c The H5F_close_all() function fails if a file cannot be closed.
1999-04-27 22:47:54 +08:00
H5I_clear_group(H5I_GROUP, FALSE);
[svn-r1169] ./configure.in ./configure [REGENERATED] ./src/H5D.c ./src/H5O.c Removed H5AC, H5B, and H5T from the default list of packages to debug (because they're pretty expensive debugging), and added H5O. Also fixed a bug for undefined variable in H5D when H5S debugging is turned on but H5T debugging is turned off. ./config/conclude.in Fixed installation of header files for building in a directory other than the source directory. This fixes a bug where H5config.h wasn't being installed. ./src/H5.c ./src/H5A.c ./src/H5D.c ./src/H5F.c ./src/H5G.c ./src/H5I.c ./src/H5Iprivate.h ./src/H5P.c ./src/H5R.c ./src/H5RA.c ./src/H5S.c ./src/H5T.c ./src/H5TB.c ./src/H5Tprivate.h ./src/H5Z.c ./src/H5detect.c ./src/H5private.h Changed the way the library shuts down again. Now it handles cycles between packages and isn't so sensitive to dependencies between packages. A package might shut down only to be restarted to process a request from some other package being shut down. Loops are detected after 100 iteractions and the shutdown is aborted with a message on standard error. This also makes it a lot easier to debug. ./src/H5A.c Fixed H5A_write() and H5A_read() so they pass a non-null background buffer to the conversion functions. This is necessary when an attribute has a compound data type. ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fsec2.c Reindented new Win32 stuff. ./src/H5Odtype.c Fixed a bug when enumeration types are used in a compound data type. The byte pointer wasn't incremented after the type information was written. ./tools/h5ls.c Compound data types display their total size because it's not always obvious from looking at the members. Scalar attributes show their space as `scalar' instead of `{}'. The index value is not printed for attributes that have only a few values. Instead the word `Data:' is printed on the first line of attribute data. Named types display their data type only if verbose output was requested.
1999-03-30 19:38:34 +08:00
} else {
/* Empty the object type table */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
for (i=0; i<H5G_ntypes_g; i++)
[svn-r1169] ./configure.in ./configure [REGENERATED] ./src/H5D.c ./src/H5O.c Removed H5AC, H5B, and H5T from the default list of packages to debug (because they're pretty expensive debugging), and added H5O. Also fixed a bug for undefined variable in H5D when H5S debugging is turned on but H5T debugging is turned off. ./config/conclude.in Fixed installation of header files for building in a directory other than the source directory. This fixes a bug where H5config.h wasn't being installed. ./src/H5.c ./src/H5A.c ./src/H5D.c ./src/H5F.c ./src/H5G.c ./src/H5I.c ./src/H5Iprivate.h ./src/H5P.c ./src/H5R.c ./src/H5RA.c ./src/H5S.c ./src/H5T.c ./src/H5TB.c ./src/H5Tprivate.h ./src/H5Z.c ./src/H5detect.c ./src/H5private.h Changed the way the library shuts down again. Now it handles cycles between packages and isn't so sensitive to dependencies between packages. A package might shut down only to be restarted to process a request from some other package being shut down. Loops are detected after 100 iteractions and the shutdown is aborted with a message on standard error. This also makes it a lot easier to debug. ./src/H5A.c Fixed H5A_write() and H5A_read() so they pass a non-null background buffer to the conversion functions. This is necessary when an attribute has a compound data type. ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fsec2.c Reindented new Win32 stuff. ./src/H5Odtype.c Fixed a bug when enumeration types are used in a compound data type. The byte pointer wasn't incremented after the type information was written. ./tools/h5ls.c Compound data types display their total size because it's not always obvious from looking at the members. Scalar attributes show their space as `scalar' instead of `{}'. The index value is not printed for attributes that have only a few values. Instead the word `Data:' is printed on the first line of attribute data. Named types display their data type only if verbose output was requested.
1999-03-30 19:38:34 +08:00
H5MM_xfree(H5G_type_g[i].desc);
H5G_ntypes_g = H5G_atypes_g = 0;
H5G_type_g = H5MM_xfree(H5G_type_g);
/* Destroy the group object id group */
H5I_destroy_group(H5I_GROUP);
[svn-r936] Changes since 19981119 ---------------------- ./src/H5.c ./src/H5A.c ./src/H5AC.c ./src/H5B.c ./src/H5D.c ./src/H5E.c ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fmpio.c ./src/H5Fsec2.c ./src/H5Fsplit.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5HG.c ./src/H5HL.c ./src/H5I.c ./src/H5Iprivate.h ./src/H5MF.c ./src/H5MM.c ./src/H5O.c ./src/H5Oattr.c ./src/H5Ocomp.c ./src/H5Ocont.c ./src/H5Odtype.c ./src/H5Oefl.c ./src/H5Ofill.c ./src/H5Olayout.c ./src/H5Omtime.c ./src/H5Oname.c ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5R.c ./src/H5RA.c ./src/H5S.c ./src/H5Sall.c ./src/H5Shyper.c ./src/H5Smpio.c ./src/H5Snone.c ./src/H5Spoint.c ./src/H5Sselect.c ./src/H5T.c ./src/H5TB.c ./src/H5Tbit.c ./src/H5Tconv.c ./src/H5V.c ./src/H5Z.c ./src/H5detect.c ./src/H5private.h Most of these changes are because the `interface_initialize_g' variable change from hbool_t to int. It's a one line change. Changed the way the library is closed so we have more control over the order the interfaces are shut down. Instead of registering an atexit() function for every interface in some haphazard order we just register one: H5_term_library() which then calls the H5*_term_interface() functions in a well-defined order. If the library is closed and then reopened repeatedly by calling H5close() and H5open() in a loop we only add one copy of the library termination functions with atexit(). Termination is a two-step process in order to help detect programming errors that would cause an infinite loop caused by the termination of one interface waking up some other previously terminated interface. The first step terminates the interface and *marks it as unusable*. After all interfaces are terminated then we mark them all as usable again. The FUNC_ENTER() macro has been modified to return failure or to dump core (depending on whether NDEBUG is defined) if we try to call an interface while it's shutting down. ./src/H5.c The H5dont_atexit() function returns failure if it's called more than once or if it's called too late. However, the error stack is not automatically printed on failure because the library might not be initialized yet ./test/chunk.c ./test/flush1.c ./test/flush2.c ./test/iopipe.c ./test/overhead.c ./test/ragged.c Changed the extra cast for Win32 so we do floating point division again -- it was just confusion about precedence and associativity of casting and the C coercion rules. Removed extra carriage returns inserted by broken operating system. ./src/H5Ffamily.c Fixed an bug where H5F_fam_write() lowered the EOF marker for one of the family members causing H5F_fam_read() to read zeros. ./test/h5test.h [NEW] ./test/h5test.c [NEW] ./test/Makefile.in ./test/bittests.c ./test/cmpd_dset.c ./test/dsets.c ./test/dtypes.c ./test/extend.c ./test/external.c Support library for test files. This isn't done yet but Katie's contractions are ~10 minutes apart so I figured I better back this stuff up just in case I'm not here next week... Eventually all test files will understand HDF5_DRIVER to name the low level file driver and parameters so we can easily test various drivers. They will also understand HDF5_PREFIX to prepend to the beginning of file names which is necessary for testing ROMIO with various drivers. Also, the cleanup function will know how to use the file name prefix and will understand different file driver naming schemes like file families. I'm not sure they'll understand the `gsf:' type prefixes yet. Note, the external test is completely commented out because I'm in the middle of modifying it. It will still compile and run but it doesn't test anything at the moment.
1998-11-21 11:36:51 +08:00
[svn-r5110] ./hdf5-devel/src/H5G.c Merge of name component length limit bugfix from 1.4 branch. A couple of very minor typos fixed.
2002-03-29 03:27:38 +08:00
/* Free the global component buffer */
H5G_comp_g = H5MM_xfree(H5G_comp_g);
H5G_comp_alloc_g = 0;
[svn-r1169] ./configure.in ./configure [REGENERATED] ./src/H5D.c ./src/H5O.c Removed H5AC, H5B, and H5T from the default list of packages to debug (because they're pretty expensive debugging), and added H5O. Also fixed a bug for undefined variable in H5D when H5S debugging is turned on but H5T debugging is turned off. ./config/conclude.in Fixed installation of header files for building in a directory other than the source directory. This fixes a bug where H5config.h wasn't being installed. ./src/H5.c ./src/H5A.c ./src/H5D.c ./src/H5F.c ./src/H5G.c ./src/H5I.c ./src/H5Iprivate.h ./src/H5P.c ./src/H5R.c ./src/H5RA.c ./src/H5S.c ./src/H5T.c ./src/H5TB.c ./src/H5Tprivate.h ./src/H5Z.c ./src/H5detect.c ./src/H5private.h Changed the way the library shuts down again. Now it handles cycles between packages and isn't so sensitive to dependencies between packages. A package might shut down only to be restarted to process a request from some other package being shut down. Loops are detected after 100 iteractions and the shutdown is aborted with a message on standard error. This also makes it a lot easier to debug. ./src/H5A.c Fixed H5A_write() and H5A_read() so they pass a non-null background buffer to the conversion functions. This is necessary when an attribute has a compound data type. ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fsec2.c Reindented new Win32 stuff. ./src/H5Odtype.c Fixed a bug when enumeration types are used in a compound data type. The byte pointer wasn't incremented after the type information was written. ./tools/h5ls.c Compound data types display their total size because it's not always obvious from looking at the members. Scalar attributes show their space as `scalar' instead of `{}'. The index value is not printed for attributes that have only a few values. Instead the word `Data:' is printed on the first line of attribute data. Named types display their data type only if verbose output was requested.
1999-03-30 19:38:34 +08:00
/* Mark closed */
interface_initialize_g = 0;
n = 1; /*H5I*/
[svn-r936] Changes since 19981119 ---------------------- ./src/H5.c ./src/H5A.c ./src/H5AC.c ./src/H5B.c ./src/H5D.c ./src/H5E.c ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fmpio.c ./src/H5Fsec2.c ./src/H5Fsplit.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5HG.c ./src/H5HL.c ./src/H5I.c ./src/H5Iprivate.h ./src/H5MF.c ./src/H5MM.c ./src/H5O.c ./src/H5Oattr.c ./src/H5Ocomp.c ./src/H5Ocont.c ./src/H5Odtype.c ./src/H5Oefl.c ./src/H5Ofill.c ./src/H5Olayout.c ./src/H5Omtime.c ./src/H5Oname.c ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5R.c ./src/H5RA.c ./src/H5S.c ./src/H5Sall.c ./src/H5Shyper.c ./src/H5Smpio.c ./src/H5Snone.c ./src/H5Spoint.c ./src/H5Sselect.c ./src/H5T.c ./src/H5TB.c ./src/H5Tbit.c ./src/H5Tconv.c ./src/H5V.c ./src/H5Z.c ./src/H5detect.c ./src/H5private.h Most of these changes are because the `interface_initialize_g' variable change from hbool_t to int. It's a one line change. Changed the way the library is closed so we have more control over the order the interfaces are shut down. Instead of registering an atexit() function for every interface in some haphazard order we just register one: H5_term_library() which then calls the H5*_term_interface() functions in a well-defined order. If the library is closed and then reopened repeatedly by calling H5close() and H5open() in a loop we only add one copy of the library termination functions with atexit(). Termination is a two-step process in order to help detect programming errors that would cause an infinite loop caused by the termination of one interface waking up some other previously terminated interface. The first step terminates the interface and *marks it as unusable*. After all interfaces are terminated then we mark them all as usable again. The FUNC_ENTER() macro has been modified to return failure or to dump core (depending on whether NDEBUG is defined) if we try to call an interface while it's shutting down. ./src/H5.c The H5dont_atexit() function returns failure if it's called more than once or if it's called too late. However, the error stack is not automatically printed on failure because the library might not be initialized yet ./test/chunk.c ./test/flush1.c ./test/flush2.c ./test/iopipe.c ./test/overhead.c ./test/ragged.c Changed the extra cast for Win32 so we do floating point division again -- it was just confusion about precedence and associativity of casting and the C coercion rules. Removed extra carriage returns inserted by broken operating system. ./src/H5Ffamily.c Fixed an bug where H5F_fam_write() lowered the EOF marker for one of the family members causing H5F_fam_read() to read zeros. ./test/h5test.h [NEW] ./test/h5test.c [NEW] ./test/Makefile.in ./test/bittests.c ./test/cmpd_dset.c ./test/dsets.c ./test/dtypes.c ./test/extend.c ./test/external.c Support library for test files. This isn't done yet but Katie's contractions are ~10 minutes apart so I figured I better back this stuff up just in case I'm not here next week... Eventually all test files will understand HDF5_DRIVER to name the low level file driver and parameters so we can easily test various drivers. They will also understand HDF5_PREFIX to prepend to the beginning of file names which is necessary for testing ROMIO with various drivers. Also, the cleanup function will know how to use the file name prefix and will understand different file driver naming schemes like file families. I'm not sure they'll understand the `gsf:' type prefixes yet. Note, the external test is completely commented out because I'm in the middle of modifying it. It will still compile and run but it doesn't test anything at the moment.
1998-11-21 11:36:51 +08:00
}
[svn-r914] Changes since 19981113 ---------------------- ./config/conclude.in ./test/Makefile.in ./tools/Makefile.in The Makefile $TESTS variable has finally been split into $TEST_PROGS and $TEST_SCRIPTS with the latter being the names of shell scripts that need to be run with `/bin/sh'. Now we don't have to copy each shell script before we run it. NOTE: THIS CHANGE DOES NOT AFFECT THE PABLO MAKEFILE SINCE THAT FILE IS A COPY OF A PREVIOUSLY GENERATED MAKEFILE. ./src/Makefile.in Added H5Snone.c to the source list. ./src/H5G.c Plugged a memory leak by emptying the object type "isa" table when the library is closed. ./src/H5Tconv.c ./src/H5Tpkg.h Added 48 new type functions to take advantage of hardware for integer type conversions. These functions are not registered in the conversion table yet because I haven't fully tested them. ./src/H5Tpkg.h Removed __unused__ qualifiers from prototypes.
1998-11-16 23:29:54 +08:00
}
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(n);
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
}
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_register_type
*
* Purpose: Register a new object type so H5G_get_type() can detect it.
* One should always register a general type before a more
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
* specific type. For instance, any object that has a datatype
* message is a datatype, but only some of those objects are
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
* datasets.
*
* Return: Success: Non-negative
*
* Failure: Negative
*
* Programmer: Robb Matzke
* Wednesday, November 4, 1998
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
[svn-r6878] Purpose: Code cleanup Description: Limit the scope on more function prototypes/macros/typedefs. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest not necessary.
2003-05-16 03:22:33 +08:00
static herr_t
[svn-r7192] Purpose: Code cleanup Description: Clean up various minor warnings in the library. Platforms tested: FreeBSD 4.8 (sleipnir) too small to need h5committest
2003-07-10 03:16:17 +08:00
H5G_register_type(H5G_obj_t type, htri_t(*isa)(H5G_entry_t*, hid_t), const char *_desc)
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
{
char *desc = NULL;
size_t i;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_register_type, FAIL);
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
assert(type>=0);
assert(isa);
assert(_desc);
/* Copy the description */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (NULL==(desc=H5MM_strdup(_desc)))
HGOTO_ERROR(H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed for object type description");
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
/*
* If the type is already registered then just update its entry without
* moving it to the end
*/
for (i=0; i<H5G_ntypes_g; i++) {
if (H5G_type_g[i].type==type) {
H5G_type_g[i].isa = isa;
H5MM_xfree(H5G_type_g[i].desc);
H5G_type_g[i].desc = desc;
[svn-r4706] Purpose: Code cleanup Description: Tweaked internal error handling macros to reduce the size of the library's object code by about 10-20%. Also cleaned up some compiler warnings... Platforms tested: FreeBSD 4.4 (sleipnir)
2001-12-13 02:40:09 +08:00
HGOTO_DONE(SUCCEED);
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
}
}
/* Increase table size */
if (H5G_ntypes_g>=H5G_atypes_g) {
size_t n = MAX(32, 2*H5G_atypes_g);
H5G_typeinfo_t *x = H5MM_realloc(H5G_type_g,
n*sizeof(H5G_typeinfo_t));
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (!x)
HGOTO_ERROR(H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed for objec type table");
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
H5G_atypes_g = n;
H5G_type_g = x;
}
/* Add a new entry */
H5G_type_g[H5G_ntypes_g].type = type;
H5G_type_g[H5G_ntypes_g].isa = isa;
H5G_type_g[H5G_ntypes_g].desc = desc; /*already copied*/
H5G_ntypes_g++;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
if (ret_value<0)
H5MM_xfree(desc);
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
}
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
/*-------------------------------------------------------------------------
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Function: H5G_component
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Purpose: Returns the pointer to the first component of the
* specified name by skipping leading slashes. Returns
* the size in characters of the component through SIZE_P not
* counting leading slashes or the null terminator.
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
[svn-r61] ./src/H5.c Changed atexit() to HDatexit(). ./src/H5AC.c Changed qsort() to HDqsort(). ./src/H5B.c Changed memcmp() to HDmemcmp(). ./src/H5E.c ./src/H5M.c Added parentheses around indirect function calls so that checkposix doesn't complain and to advertise that it's an indirect function call. ./src/H5F.c ./src/H5G.c Added errors field to each function prologue to test the script mentioned above. ./src/H5H.c Changed fputc() to HDfputc(). ./src/H5MM.c ./src/H5MMprivate.h The argument for H5MM_xfree() is a pointer to a constant even though the argument to free() isn't. This reduces the number of warnings in other parts of the code but creates a new warning in this file. ./src/H5O.c Plugged a few memory leaks that happen during error handling. ./src/H5Oname.c ./src/debug.c Added `HD' to the beginning of some posix functions.
1997-09-02 23:38:26 +08:00
* Errors:
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Return: Success: Ptr into NAME.
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Failure: Ptr to the null terminator of NAME.
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Programmer: Robb Matzke
* matzke@llnl.gov
* Aug 11 1997
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
static const char *
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
H5G_component(const char *name, size_t *size_p)
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
{
[svn-r5467] Purpose: Code cleanup. Description: Took Robb's recent ideas for improving the FUNC_ENTER/FUNC_LEAVE macros equivalents in the SAF library and adapted them to our library. I added an additional macro which is equivalent to FUNC_ENTER: FUNC_ENTER_NOINIT - Has the API tracing code, etc. from FUNC_ENTER but none of the library or interface initialization code. This is to be used _only_ for static functions and those which explicitly cannot have the library or interface initialization code enabled (like the API termination routines, etc.). This allowed many more of the functions in the library [but not all yet :-(] to be wrapped with FUNC_ENTER[_NOINIT]/FUNC_LEAVE pairs. It also reduced the size of the library and executables (by cutting out a bunch of code which was never executed), I'll e-mail the exact results when I've finished editing it. Platforms tested: IRIX64 6.5 (modi4)
2002-05-29 02:17:12 +08:00
/* Use FUNC_ENTER_NOINIT here to avoid performance issues */
FUNC_ENTER_NOINIT(H5G_component);
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
assert(name);
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
while ('/' == *name)
name++;
if (size_p)
*size_p = HDstrcspn(name, "/");
[svn-r5467] Purpose: Code cleanup. Description: Took Robb's recent ideas for improving the FUNC_ENTER/FUNC_LEAVE macros equivalents in the SAF library and adapted them to our library. I added an additional macro which is equivalent to FUNC_ENTER: FUNC_ENTER_NOINIT - Has the API tracing code, etc. from FUNC_ENTER but none of the library or interface initialization code. This is to be used _only_ for static functions and those which explicitly cannot have the library or interface initialization code enabled (like the API termination routines, etc.). This allowed many more of the functions in the library [but not all yet :-(] to be wrapped with FUNC_ENTER[_NOINIT]/FUNC_LEAVE pairs. It also reduced the size of the library and executables (by cutting out a bunch of code which was never executed), I'll e-mail the exact results when I've finished editing it. Platforms tested: IRIX64 6.5 (modi4)
2002-05-29 02:17:12 +08:00
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(name);
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
}
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_basename
*
* Purpose: Returns a pointer to the last component of the specified
* name. The length of the component is returned through SIZE_P.
* The base name is followed by zero or more slashes and a null
* terminator, but SIZE_P does not count the slashes or the null
* terminator.
*
* Note: The base name of the root directory is a single slash.
*
* Return: Success: Ptr to base name.
*
* Failure: Ptr to the null terminator.
*
* Programmer: Robb Matzke
* Thursday, September 17, 1998
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
static const char *
H5G_basename(const char *name, size_t *size_p)
{
[svn-r1101] Changes since 19990219 ---------------------- ./config/irix64 The old (-32) compiler is now supported by setting envrionment CC='cc -32'. The -64 compiler is the default or you can set CC='cc -64'. ./src/H5A.c ./src/H5D.c ./src/H5F.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5G.c ./src/H5I.c ./src/H5Ocomp.c ./src/H5P.c ./src/H5R.c ./src/H5RA.c ./src/H5T.c ./src/H5Tbit.c ./src/H5Tconv.c ./src/H5Z.c ./src/H5detect.c ./test/big.c ./test/cmpd_dset.c ./test/dsets.c ./test/dtypes.c ./test/enum.c ./test/mtime.c ./test/ohdr.c ./tools/h5ls.c Fixed lots of warnings on Irix64. Mailed a few remaining warnings in H5S to Quincey and a few in the dumper to Ruey-Hsia.
1999-02-25 23:40:27 +08:00
size_t i;
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
[svn-r5467] Purpose: Code cleanup. Description: Took Robb's recent ideas for improving the FUNC_ENTER/FUNC_LEAVE macros equivalents in the SAF library and adapted them to our library. I added an additional macro which is equivalent to FUNC_ENTER: FUNC_ENTER_NOINIT - Has the API tracing code, etc. from FUNC_ENTER but none of the library or interface initialization code. This is to be used _only_ for static functions and those which explicitly cannot have the library or interface initialization code enabled (like the API termination routines, etc.). This allowed many more of the functions in the library [but not all yet :-(] to be wrapped with FUNC_ENTER[_NOINIT]/FUNC_LEAVE pairs. It also reduced the size of the library and executables (by cutting out a bunch of code which was never executed), I'll e-mail the exact results when I've finished editing it. Platforms tested: IRIX64 6.5 (modi4)
2002-05-29 02:17:12 +08:00
FUNC_ENTER_NOINIT(H5G_basename);
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
/* Find the end of the base name */
[svn-r4022] Purpose: Code Cleanup Description: Changed POSIX function calls to the HD* way of doing things. Checked with checkposix and tested on Linux. Platforms tested: Linux
2001-06-20 05:49:01 +08:00
i = HDstrlen(name);
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
while (i>0 && '/'==name[i-1])
--i;
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
/* Skip backward over base name */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
while (i>0 && '/'!=name[i-1])
--i;
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
/* Watch out for root special case */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if ('/'==name[i] && size_p)
*size_p = 1;
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(name+i);
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
}
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_normalize
*
* Purpose: Returns a pointer to a new string which has duplicate and
* trailing slashes removed from it.
*
* Return: Success: Ptr to normalized name.
* Failure: NULL
*
* Programmer: Quincey Koziol
* Saturday, August 16, 2003
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
static char *
H5G_normalize(const char *name)
{
char *norm; /* Pointer to the normalized string */
size_t s,d; /* Positions within the strings */
unsigned last_slash; /* Flag to indicate last character was a slash */
char *ret_value; /* Return value */
FUNC_ENTER_NOINIT(H5G_normalize);
/* Sanity check */
assert(name);
/* Duplicate the name, to return */
if (NULL==(norm=H5MM_strdup(name)))
HGOTO_ERROR(H5E_RESOURCE, H5E_NOSPACE, NULL, "memory allocation failed for normalized string");
/* Walk through the characters, omitting duplicated '/'s */
s=d=0;
last_slash=0;
while(name[s]!='\0') {
if(name[s]=='/')
if(last_slash)
;
else {
norm[d++]=name[s];
last_slash=1;
} /* end else */
else {
norm[d++]=name[s];
last_slash=0;
} /* end else */
s++;
} /* end while */
/* Terminate normalized string */
norm[d]='\0';
/* Check for final '/' on normalized name & eliminate it */
if(d>1 && last_slash)
norm[d-1]='\0';
/* Set return value */
ret_value=norm;
done:
FUNC_LEAVE_NOAPI(ret_value);
} /* end H5G_normalize() */
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
/*-------------------------------------------------------------------------
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Function: H5G_namei
*
* Purpose: Translates a name to a symbol table entry.
*
* If the specified name can be fully resolved, then this
* function returns the symbol table entry for the named object
* through the OBJ_ENT argument. The symbol table entry for the
* group containing the named object is returned through the
* GRP_ENT argument if it is non-null. However, if the name
* refers to the root object then the GRP_ENT will be
* initialized with an undefined object header address. The
* REST argument, if present, will point to the null terminator
* of NAME.
*
* If the specified name cannot be fully resolved, then OBJ_ENT
* is initialized with the undefined object header address. The
* REST argument will point into the NAME argument to the start
* of the component that could not be located. The GRP_ENT will
* contain the entry for the symbol table that was being
* searched at the time of the failure and will have an
* undefined object header address if the search failed at the
* root object. For instance, if NAME is `/foo/bar/baz' and the
* root directory exists and contains an entry for `foo', and
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
* foo is a group that contains an entry for bar, but bar is not
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* a group, then the results will be that REST points to `baz',
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
* OBJ_ENT has an undefined object header address, and GRP_ENT
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* is the symbol table entry for `bar' in `/foo'.
*
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
* Every file has a root group whose name is `/'. Components of
* a name are separated from one another by one or more slashes
* (/). Slashes at the end of a name are ignored. If the name
* begins with a slash then the search begins at the root group
* of the file containing LOC_ENT. Otherwise it begins at
* LOC_ENT. The component `.' is a no-op, but `..' is not
* understood by this function (unless it appears as an entry in
* the symbol table).
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
*
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
* Symbolic links are followed automatically, but if TARGET
* includes the H5G_TARGET_SLINK bit and the last component of
* the name is a symbolic link then that link is not followed.
* The *NLINKS value is decremented each time a link is followed
* and link traversal fails if the value would become negative.
* If NLINKS is the null pointer then a default value is used.
[svn-r745] Changes since 19981002 ---------------------- ./doc/html/H5.format.html ./src/H5HG.c Fixed a bug in the global heap that caused H5HG_read() to write past the end of the buffer in certain cases. ./test/big.c The test is skipped if hdf5 was configured with `--disable-hsizet'. ./src/H5Ofill.c Data type conversions are implemented for the fill value. ./src/H5.c Tracing prints one of H5P_FILE_CREATE, H5P_FILE_ACCESS, H5P_DATASET_CREATE, H5P_DATASET_XFER, or H5P_MOUNT instead of the more cryptic H5I_TEMPLATE_* constants. ./src/H5D.c Removed prototype for H5D_find_name(). ./src/H5I.c The GROUP_MASK and ID_MASK are both calculated from GROUP_BITS instead of being set by hand. We don't use the sign bit of hid_t; all valid hid_t values are positive so we can say things like `if ((file=H5Fopen(...))<0)'. Changed `(int)pow(2.0,x)' to `1<<x' so we don't have to worry about rounding. Fixed H5I_get_type() so it doesn't always fail an assertion. ./src/H5E.c ./src/H5Epublic.h Added minor error H5E_MOUNT ./src/H5F.c ./src/H5Fprivate.h Added H5Fmount() and H5Funmount(). Mounting and unmounting works as documented but some of the other things aren't implemented yet, the biggest being current working groups always acting on the root of the mount tree, and H5Fclose() closing the entire tree. The rest of the stuff will be added shortly... ./src/H5P.c ./src/H5Ppublic.h Added the H5P_MOUNT property list but haven't implemented any particular properties for it yet. ./src/H5Gstab.c Hard links across files return an error instead of failing an assertion.
1998-10-09 01:13:14 +08:00
*
* Mounted files are handled by calling H5F_mountpoint() after
* each step of the translation. If the input argument to that
* function is a mount point then the argument shall be replaced
* with information about the root group of the mounted file.
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
* But if TARGET includes the H5G_TARGET_MOUNT bit and the last
* component of the name is a mount point then H5F_mountpoint()
* is not called and information about the mount point itself is
* returned.
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
*
[svn-r61] ./src/H5.c Changed atexit() to HDatexit(). ./src/H5AC.c Changed qsort() to HDqsort(). ./src/H5B.c Changed memcmp() to HDmemcmp(). ./src/H5E.c ./src/H5M.c Added parentheses around indirect function calls so that checkposix doesn't complain and to advertise that it's an indirect function call. ./src/H5F.c ./src/H5G.c Added errors field to each function prologue to test the script mentioned above. ./src/H5H.c Changed fputc() to HDfputc(). ./src/H5MM.c ./src/H5MMprivate.h The argument for H5MM_xfree() is a pointer to a constant even though the argument to free() isn't. This reduces the number of warnings in other parts of the code but creates a new warning in this file. ./src/H5O.c Plugged a few memory leaks that happen during error handling. ./src/H5Oname.c ./src/debug.c Added `HD' to the beginning of some posix functions.
1997-09-02 23:38:26 +08:00
* Errors:
*
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
* Return: Success: Non-negative if name can be fully resolved.
* See above for values of REST, GRP_ENT, and
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
* OBJ_ENT. NLINKS has been decremented for
* each symbolic link that was followed.
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
* Failure: Negative if the name could not be fully
* resolved. See above for values of REST,
* GRP_ENT, and OBJ_ENT.
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Programmer: Robb Matzke
* matzke@llnl.gov
* Aug 11 1997
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
* Modifications:
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Robb Matzke, 2002-03-28
* The component name buffer on the stack has been replaced by
* a dynamically allocated buffer on the heap in order to
* remove limitations on the length of a name component.
* There are two reasons that the buffer pointer is global:
* (1) We want to be able to reuse the buffer without
* allocating and freeing it each time this function is
* called.
* (2) We need to be able to free it from H5G_term_interface()
* when the library terminates.
*
* Pedro Vicente, <pvn@ncsa.uiuc.edu> 22 Aug 2002
* Modified to deep copies of symbol table entries
* Added `id to name' support.
*
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
* Quincey Koziol, 2003-01-06
* Added "action" and "ent" parameters to allow different actions when
* working on the last component of a name. (Specifically, this allows
* inserting an entry into a group, instead of trying to look it up)
*
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*-------------------------------------------------------------------------
*/
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
static herr_t
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
H5G_namei(H5G_entry_t *loc_ent, const char *name, const char **rest/*out*/,
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
H5G_entry_t *grp_ent/*out*/, H5G_entry_t *obj_ent/*out*/,
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
unsigned target, int *nlinks/*out*/, H5G_namei_act_t action,
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
H5G_entry_t *ent, hid_t dxpl_id)
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
{
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
H5G_entry_t _grp_ent; /*entry for current group */
H5G_entry_t _obj_ent; /*entry found */
size_t nchars; /*component name length */
int _nlinks = H5G_NLINKS;
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
const char *s = NULL;
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
unsigned null_obj; /* Flag to indicate this function was called with obj_ent set to NULL */
unsigned null_grp; /* Flag to indicate this function was called with grp_ent set to NULL */
unsigned group_copy = 0; /* Flag to indicate that the group entry is copied */
unsigned last_comp = 0; /* Flag to indicate that a component is the last component in the name */
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r5467] Purpose: Code cleanup. Description: Took Robb's recent ideas for improving the FUNC_ENTER/FUNC_LEAVE macros equivalents in the SAF library and adapted them to our library. I added an additional macro which is equivalent to FUNC_ENTER: FUNC_ENTER_NOINIT - Has the API tracing code, etc. from FUNC_ENTER but none of the library or interface initialization code. This is to be used _only_ for static functions and those which explicitly cannot have the library or interface initialization code enabled (like the API termination routines, etc.). This allowed many more of the functions in the library [but not all yet :-(] to be wrapped with FUNC_ENTER[_NOINIT]/FUNC_LEAVE pairs. It also reduced the size of the library and executables (by cutting out a bunch of code which was never executed), I'll e-mail the exact results when I've finished editing it. Platforms tested: IRIX64 6.5 (modi4)
2002-05-29 02:17:12 +08:00
FUNC_ENTER_NOINIT(H5G_namei);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Set up "out" parameters */
if (rest)
*rest = name;
if (!grp_ent) {
grp_ent = &_grp_ent;
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
null_grp = 1;
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
} /* end if */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
else
null_grp = 0;
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
if (!obj_ent) {
obj_ent = &_obj_ent;
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
null_obj = 1;
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
} /* end if */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
else
null_obj = 0;
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
if (!nlinks)
nlinks = &_nlinks;
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Check args */
if (!name || !*name)
HGOTO_ERROR (H5E_SYM, H5E_NOTFOUND, FAIL, "no name given");
if (!loc_ent)
HGOTO_ERROR (H5E_SYM, H5E_NOTFOUND, FAIL, "no current working group");
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
/*
* Where does the searching start? For absolute names it starts at the
* root of the file; for relative names it starts at CWG.
*/
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Check if we need to get the root group's entry */
if ('/' == *name) {
H5G_t *tmp_grp; /* Temporary pointer to root group of file */
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
tmp_grp=H5G_rootof(loc_ent->file);
assert(tmp_grp);
/* Set the location entry to the root group's entry*/
loc_ent=&(tmp_grp->ent);
} /* end if */
/* Deep copy of the symbol table entry */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
if (H5G_ent_copy(obj_ent, loc_ent,H5G_COPY_DEEP)<0)
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
HGOTO_ERROR(H5E_DATATYPE, H5E_CANTOPENOBJ, FAIL, "unable to copy entry");
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r673] Changes since 19980908 ---------------------- ./COPYING Reformatted as text instead of C. Removed zlib crew from the list of contributors since no zlib code is actually in the hdf5 library. ./INSTALL ./INSTALL.ascired ./MANIFEST Minor updates for Beta release including version number change. ./INSTALL_MAINT Added information about making a release. ./RELEASE Updated function list based on public header files. ./bin/checkposix Got rid of complaints about some obvious things. ./doc/html/H5.api.html ./doc/html/RM_H5F.html ./src/H5F.c ./src/H5Fpublic.h ./test/tfile.c Changed H5Fget_create_template() and H5Fget_access_template() to H5Fget_create_plist() and H5Fget_access_plist() since that conforms better to lots of other names. ./doc/html/Datatypes.html ./doc/html/ExternalFiles.html ./doc/html/Files.html ./doc/html/H5.api.html ./doc/html/H5.sample_code.html ./doc/html/RM_H5F.html ./doc/html/RM_H5Front.html Changed `template' to `property list', etc. ./doc/html/Ragged.html [NEW] Documentation for ragged arrays. ./src/H5Iprivate.h ./src/H5Ipublic.h ./src/H5I.c Changed the scope of some symbols to be more local. ./src/H5.c ./src/H5AC.c ./src/H5D.c ./src/H5E.c ./src/H5F.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fsec2.c ./src/H5Fsplit.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gnode.c ./src/H5HG.c ./src/H5I.c ./src/H5O.c ./src/H5Ocomp.c ./src/H5Odtype.c ./src/H5Oefl.c ./src/H5Omtime.c ./src/H5Oname.c ./src/H5P.c ./src/H5S.c ./src/H5Shyper.c ./src/H5Tbit.c ./src/H5Tconv.c ./src/H5V.c ./src/H5Z.c ./src/H5private.h Fixed some violations of our naming scheme by adding HD to the beginning of all Posix functions.
1998-09-09 03:15:44 +08:00
HDmemset(grp_ent, 0, sizeof(H5G_entry_t));
[svn-r1568] Changes since 19990730 ---------------------- This extensive change is the virtual file layer implementation. I've ported and tested the sec2, family, and core drivers and only ported the mpio driver (Albert will test it). So if you need MPIO I would recommend sticking with the previous version for a while. You will get a few compile warnings about split and stdio drivers not being implemented and possibly tracing information not inserted in some of the drivers. You can safely ignore them but I plan to fix them. I'm still working on the split driver because I just realized that it needs a part of the VFL that isn't written yet. Documentation is being updated also because there were some minor changes (mostly just name changes). It should be available on my web site later this week. ./MANIFEST ./src/Makefile.in ./src/hdf5.h ./src/H5Flow.c [REMOVED] ./src/H5Fstdio.c [REMOVED] ./src/H5Fsec2.c [REMOVED] ./src/H5Fsplit.c [REMOVED] ./src/H5Fmpio.c [REMOVED] ./src/H5Ffamily.c [REMOVED] ./src/H5Fcore.c [REMOVED] ./src/H5MFpublic.h [REMOVED] ./src/H5FD.c [NEW] ./src/H5FDcore.c [NEW] ./src/H5FDcore.h [NEW] ./src/H5FDfamily.c [NEW] ./src/H5FDfamily.h [NEW] ./src/H5FDmpio.c [NEW] ./src/H5FDmpio.h [NEW] ./src/H5FDprivate.h [NEW] ./src/H5FDpublic.h [NEW] ./src/H5FDsec2.c [NEW] ./src/H5FDsec2.h [NEW] Removed/added files for virtual file layer. ./bin/trace ./src/H5.c Removed unused public datatypes and added new VFL public datatypes. Changed an error message. ./config/BlankForm ./config/dec-flags ./config/gnu-flags ./config/hpux10.20 ./config/hpux9.03 ./config/irix5.x ./config/irix6.x ./config/solaris2.x ./config/unicosmk Removed the H5F_OPT_SEEK and H5F_LOW_DFLT constants from the configuration since they're no longer applicable. The default file driver is always the sec2 driver and it always optimizes calls to lseek() or lseek64(). ./config/depend.in C preprocessor errors generated during automatic dependency building are sent to /dev/null to prevent them from appearing twice in the make output. ./src/H5AC.c ./src/H5B.c ./src/H5D.c ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5HG.c ./src/H5HL.c ./src/H5O.c ./src/H5Oattr.c ./src/H5Odtype.c ./src/H5Oefl.c ./src/H5Oshared.c ./src/H5T.c ./src/H5detect.c ./test/ohdr.c Changed H5F_ADDR_UNDEF to HADDR_UNDEF to be more consistent with the `haddr_t' datatype which is now a public type. ./src/H5D.c ./src/H5P.c ./src/H5Ppublic.h ./src/H5Tconv.c ./test/cmpd_dset.c ./test/dsets.c ./test/overhead.c ./test/tselect.c ./test/tvltypes.c The H5P_DATASET_XFER constant was changed to H5P_DATA_XFER because the properties apply to all types of I/O operations, not just datasets. ./src/H5B.c ./src/H5Bprivate.h ./src/H5D.c ./src/H5Dpublic.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fistore.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5HG.c ./src/H5HL.c ./src/H5O.c ./src/H5R.c ./src/H5Sall.c ./src/H5Shyper.c ./src/H5Smpio.c ./src/H5Spoint.c ./src/H5Sprivate.h ./test/big.c ./test/h5test.c ./test/istore.c ./testpar/t_dset.c ./testpar/t_file.c ./tools/h5debug.c ./tools/h5ls.c Modified to work with the virtual file layer by calling H5FD_* functions instead of H5F_low_* functions and by passing file access and data transfer properties by object ID instead of pointer. Changed H5D_transfer_t to H5FD_mpio_xfer_t since the COLLECTIVE vs. INDEPENDENT transfer mode is specific to the MPIO file driver. Moved MPIO-specific stuff into the MPIO driver. ./src/H5B.c ./src/H5D.c ./src/H5Fprivate.h The H5F_mpio_* private functions were renamed and placed in the H5FDmpio driver except those which appeared in H5Smpio.c. ./src/H5E.c ./src/H5Epublic.h Added major error number H5E_VFL for virtual file layer related errors. ./src/H5F.c ./src/H5Fprivate.h Changed the logic that controls whether the boot block is written. Instead of assuming that the first call to write the boot block is only to allocate space, I've added a function argument which makes this explicit. Changed the way files are compared so that a driver-defined comparison function can be called. Files which belong to different drivers are always considered different. Removed H5F_driver_t since file drivers are now identified by object ID instead of a special non-user-extendible datatype. Removed all the hard-coded low-level file properties which have been replaced by the various file drivers. ./src/H5I.c ./src/H5Iprivate.h Added the H5I_inc_ref() which was removed a few months ago since we finally have a use for it. ./src/H5Ipublic.h Added the H5I_VFL object ID type to identify file drivers in the virtual file layer. ./src/H5MF.c ./src/H5MFprivate.h Moved all the allocation/deallocation code into the virtual file layer which allows file drivers to override much of it. ./src/H5P.c ./src/H5Ppublic.h Moved file driver-specific code into the various file driver files. The H5Pcopy() and H5Pclose() functions make calls into the virtual file driver to manage the memory for driver-specific file access and data transfer properties. ./src/H5private.h ./src/H5public.h The `haddr_t' type is now public. ./test/tfile.c Added a few more comments.
1999-08-11 04:21:32 +08:00
grp_ent->header = HADDR_UNDEF;
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/* traverse the name */
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
while ((name = H5G_component(name, &nchars)) && *name) {
/* Update the "rest of name" pointer */
if (rest)
*rest = name;
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
/*
* Copy the component name into a null-terminated buffer so
* we can pass it down to the other symbol table functions.
*/
[svn-r5110] ./hdf5-devel/src/H5G.c Merge of name component length limit bugfix from 1.4 branch. A couple of very minor typos fixed.
2002-03-29 03:27:38 +08:00
if (nchars+1 > H5G_comp_alloc_g) {
H5G_comp_alloc_g = MAX3(1024, 2*H5G_comp_alloc_g, nchars+1);
H5G_comp_g = H5MM_realloc(H5G_comp_g, H5G_comp_alloc_g);
if (!H5G_comp_g) {
H5G_comp_alloc_g = 0;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_NOSPACE, FAIL, "unable to allocate component buffer");
[svn-r5110] ./hdf5-devel/src/H5G.c Merge of name component length limit bugfix from 1.4 branch. A couple of very minor typos fixed.
2002-03-29 03:27:38 +08:00
}
}
HDmemcpy(H5G_comp_g, name, nchars);
H5G_comp_g[nchars] = '\0';
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
/*
* The special name `.' is a no-op.
*/
[svn-r5110] ./hdf5-devel/src/H5G.c Merge of name component length limit bugfix from 1.4 branch. A couple of very minor typos fixed.
2002-03-29 03:27:38 +08:00
if ('.' == H5G_comp_g[0] && !H5G_comp_g[1]) {
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
name += nchars;
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
continue;
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
}
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
/*
* Advance to the next component of the name.
*/
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* If we've already copied a new entry into the group entry,
* it needs to be freed before overwriting it with another entry
*/
if(group_copy)
H5G_free_ent_name(grp_ent);
/* Transfer "ownership" of the entry's information to the group entry */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
H5G_ent_copy(grp_ent,obj_ent,H5G_COPY_SHALLOW);
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
HDmemset(obj_ent, 0, sizeof(H5G_entry_t));
[svn-r1568] Changes since 19990730 ---------------------- This extensive change is the virtual file layer implementation. I've ported and tested the sec2, family, and core drivers and only ported the mpio driver (Albert will test it). So if you need MPIO I would recommend sticking with the previous version for a while. You will get a few compile warnings about split and stdio drivers not being implemented and possibly tracing information not inserted in some of the drivers. You can safely ignore them but I plan to fix them. I'm still working on the split driver because I just realized that it needs a part of the VFL that isn't written yet. Documentation is being updated also because there were some minor changes (mostly just name changes). It should be available on my web site later this week. ./MANIFEST ./src/Makefile.in ./src/hdf5.h ./src/H5Flow.c [REMOVED] ./src/H5Fstdio.c [REMOVED] ./src/H5Fsec2.c [REMOVED] ./src/H5Fsplit.c [REMOVED] ./src/H5Fmpio.c [REMOVED] ./src/H5Ffamily.c [REMOVED] ./src/H5Fcore.c [REMOVED] ./src/H5MFpublic.h [REMOVED] ./src/H5FD.c [NEW] ./src/H5FDcore.c [NEW] ./src/H5FDcore.h [NEW] ./src/H5FDfamily.c [NEW] ./src/H5FDfamily.h [NEW] ./src/H5FDmpio.c [NEW] ./src/H5FDmpio.h [NEW] ./src/H5FDprivate.h [NEW] ./src/H5FDpublic.h [NEW] ./src/H5FDsec2.c [NEW] ./src/H5FDsec2.h [NEW] Removed/added files for virtual file layer. ./bin/trace ./src/H5.c Removed unused public datatypes and added new VFL public datatypes. Changed an error message. ./config/BlankForm ./config/dec-flags ./config/gnu-flags ./config/hpux10.20 ./config/hpux9.03 ./config/irix5.x ./config/irix6.x ./config/solaris2.x ./config/unicosmk Removed the H5F_OPT_SEEK and H5F_LOW_DFLT constants from the configuration since they're no longer applicable. The default file driver is always the sec2 driver and it always optimizes calls to lseek() or lseek64(). ./config/depend.in C preprocessor errors generated during automatic dependency building are sent to /dev/null to prevent them from appearing twice in the make output. ./src/H5AC.c ./src/H5B.c ./src/H5D.c ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5HG.c ./src/H5HL.c ./src/H5O.c ./src/H5Oattr.c ./src/H5Odtype.c ./src/H5Oefl.c ./src/H5Oshared.c ./src/H5T.c ./src/H5detect.c ./test/ohdr.c Changed H5F_ADDR_UNDEF to HADDR_UNDEF to be more consistent with the `haddr_t' datatype which is now a public type. ./src/H5D.c ./src/H5P.c ./src/H5Ppublic.h ./src/H5Tconv.c ./test/cmpd_dset.c ./test/dsets.c ./test/overhead.c ./test/tselect.c ./test/tvltypes.c The H5P_DATASET_XFER constant was changed to H5P_DATA_XFER because the properties apply to all types of I/O operations, not just datasets. ./src/H5B.c ./src/H5Bprivate.h ./src/H5D.c ./src/H5Dpublic.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fistore.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5HG.c ./src/H5HL.c ./src/H5O.c ./src/H5R.c ./src/H5Sall.c ./src/H5Shyper.c ./src/H5Smpio.c ./src/H5Spoint.c ./src/H5Sprivate.h ./test/big.c ./test/h5test.c ./test/istore.c ./testpar/t_dset.c ./testpar/t_file.c ./tools/h5debug.c ./tools/h5ls.c Modified to work with the virtual file layer by calling H5FD_* functions instead of H5F_low_* functions and by passing file access and data transfer properties by object ID instead of pointer. Changed H5D_transfer_t to H5FD_mpio_xfer_t since the COLLECTIVE vs. INDEPENDENT transfer mode is specific to the MPIO file driver. Moved MPIO-specific stuff into the MPIO driver. ./src/H5B.c ./src/H5D.c ./src/H5Fprivate.h The H5F_mpio_* private functions were renamed and placed in the H5FDmpio driver except those which appeared in H5Smpio.c. ./src/H5E.c ./src/H5Epublic.h Added major error number H5E_VFL for virtual file layer related errors. ./src/H5F.c ./src/H5Fprivate.h Changed the logic that controls whether the boot block is written. Instead of assuming that the first call to write the boot block is only to allocate space, I've added a function argument which makes this explicit. Changed the way files are compared so that a driver-defined comparison function can be called. Files which belong to different drivers are always considered different. Removed H5F_driver_t since file drivers are now identified by object ID instead of a special non-user-extendible datatype. Removed all the hard-coded low-level file properties which have been replaced by the various file drivers. ./src/H5I.c ./src/H5Iprivate.h Added the H5I_inc_ref() which was removed a few months ago since we finally have a use for it. ./src/H5Ipublic.h Added the H5I_VFL object ID type to identify file drivers in the virtual file layer. ./src/H5MF.c ./src/H5MFprivate.h Moved all the allocation/deallocation code into the virtual file layer which allows file drivers to override much of it. ./src/H5P.c ./src/H5Ppublic.h Moved file driver-specific code into the various file driver files. The H5Pcopy() and H5Pclose() functions make calls into the virtual file driver to manage the memory for driver-specific file access and data transfer properties. ./src/H5private.h ./src/H5public.h The `haddr_t' type is now public. ./test/tfile.c Added a few more comments.
1999-08-11 04:21:32 +08:00
obj_ent->header = HADDR_UNDEF;
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Set flag that we've copied a new entry into the group entry */
group_copy =1;
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/* Check if this is the last component of the name */
if(!((s=H5G_component(name+nchars, NULL)) && *s))
last_comp=1;
switch(action) {
case H5G_NAMEI_TRAVERSE:
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_stab_find(grp_ent, H5G_comp_g, obj_ent/*out*/, dxpl_id )<0) {
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/*
* Component was not found in the current symbol table, possibly
* because GRP_ENT isn't a symbol table.
*/
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, FAIL, "component not found");
}
break;
case H5G_NAMEI_INSERT:
if(!last_comp) {
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_stab_find(grp_ent, H5G_comp_g, obj_ent/*out*/, dxpl_id )<0) {
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/*
* Component was not found in the current symbol table, possibly
* because GRP_ENT isn't a symbol table.
*/
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, FAIL, "component not found");
}
} /* end if */
else {
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_stab_insert(grp_ent, H5G_comp_g, ent, dxpl_id) < 0)
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINSERT, FAIL, "unable to insert name");
HGOTO_DONE(SUCCEED);
} /* end else */
break;
} /* end switch */
[svn-r5907] Purpose: bug turnaround in test_hdf5_fortran on a h5fmount_c call Description: the H5I_nmembers(H5I_GROUP) call that is made inside h5fmount_c -> H5G_replace_name -> H5I_nmembers(H5I_GROUP) is not detecteting correctly that the entry belongs to the group list therefore a call is made to H5I_search(H5I_DATATYPE) (there are always non-named datatypes) and H5G_replace_ent is called with a search for the datatype group ent = H5T_entof((H5T_t*)obj_ptr); which returns a NULL ent a tweak for this is to add if( !ent) goto done; Platforms tested: windows 2000, Linux, Solaris with fortran
2002-08-31 05:59:22 +08:00
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/*
* If we found a symbolic link then we should follow it. But if this
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
* is the last component of the name and the H5G_TARGET_SLINK bit of
* TARGET is set then we don't follow it.
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
*/
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(H5G_CACHED_SLINK==obj_ent->type &&
(0==(target & H5G_TARGET_SLINK) || !last_comp)) {
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if ((*nlinks)-- <= 0)
HGOTO_ERROR (H5E_SYM, H5E_SLINK, FAIL, "too many symbolic links");
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_traverse_slink (grp_ent, obj_ent, nlinks, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_NOTFOUND, FAIL, "symbolic link traversal failed");
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
}
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
/*
* Resolve mount points to the mounted group. Do not do this step if
* the H5G_TARGET_MOUNT bit of TARGET is set and this is the last
* component of the name.
*/
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if (0==(target & H5G_TARGET_MOUNT) || !last_comp)
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
H5F_mountpoint(obj_ent/*in,out*/);
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
/* next component */
name += nchars;
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
} /* end while */
[svn-r5907] Purpose: bug turnaround in test_hdf5_fortran on a h5fmount_c call Description: the H5I_nmembers(H5I_GROUP) call that is made inside h5fmount_c -> H5G_replace_name -> H5I_nmembers(H5I_GROUP) is not detecteting correctly that the entry belongs to the group list therefore a call is made to H5I_search(H5I_DATATYPE) (there are always non-named datatypes) and H5G_replace_ent is called with a search for the datatype group ent = H5T_entof((H5T_t*)obj_ptr); which returns a NULL ent a tweak for this is to add if( !ent) goto done; Platforms tested: windows 2000, Linux, Solaris with fortran
2002-08-31 05:59:22 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Update the "rest of name" pointer */
if (rest)
*rest = name; /*final null */
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* If we started with a NULL obj_ent, free the entry information */
if(null_obj)
H5G_free_ent_name(obj_ent);
/* If we started with a NULL grp_ent and we copied something into it, free the entry information */
if(null_grp && group_copy)
H5G_free_ent_name(grp_ent);
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
}
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_traverse_slink
*
* Purpose: Traverses symbolic link. The link head appears in the group
* whose entry is GRP_ENT and the link head entry is OBJ_ENT.
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Return: Success: Non-negative, OBJ_ENT will contain information
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
* about the object to which the link points and
* GRP_ENT will contain the information about
* the group in which the link tail appears.
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Failure: Negative
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
*
* Programmer: Robb Matzke
* Friday, April 10, 1998
*
* Modifications:
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
*
* Pedro Vicente, <pvn@ncsa.uiuc.edu> 22 Aug 2002
* Added `id to name' support.
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
*
*-------------------------------------------------------------------------
*/
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
static herr_t
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
H5G_traverse_slink (H5G_entry_t *grp_ent/*in,out*/,
H5G_entry_t *obj_ent/*in,out*/,
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
int *nlinks/*in,out*/, hid_t dxpl_id)
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
{
H5O_stab_t stab_mesg; /*info about local heap */
const char *clv = NULL; /*cached link value */
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
char *linkval = NULL; /*the copied link value */
H5G_entry_t tmp_grp_ent; /* Temporary copy of group entry */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
H5RS_str_t *tmp_user_path_r=NULL, *tmp_canon_path_r=NULL; /* Temporary pointer to object's user path & canonical path */
[svn-r7498] Purpose: Code cleanup Description: Clean up various warnings and parameter mis-matches, etc. Platforms tested: FreeBSD 4.9 (sleipnir) too minor to need h5committest
2003-09-20 04:13:47 +08:00
const H5HL_t *heap;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_traverse_slink, FAIL);
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Portably initialize the temporary group entry */
HDmemset(&tmp_grp_ent,0,sizeof(H5G_entry_t));
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/* Get the link value */
[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
if (NULL==H5O_read (grp_ent, H5O_STAB_ID, 0, &stab_mesg, dxpl_id))
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_NOTFOUND, FAIL, "unable to determine local heap address");
[svn-r7489] Purpose: Update Description: A lot of modifications for the FPHDF5 stuff: H5AC.c H5ACprivate.h - Removed AC_find (it's replaced with AC_protect and AC_unprotect). Added flushing if it's an FPHDF5 driver and we're doing an AC_set or AC_unprotect with the dirty flag set. H5B.c - Split up the B_flush function into different functions since the one function was doing serialization which is better left as a separate entity. H5D.c - Removed some FPHDF5 code that was incorrect H5F.c - Split up the F_flush function so that it no longer allocates file space. Created new functions (F_init_superblock, F_read_superblock, and F_write_superblock) for greater modularity and so that the FPHDF5 non-captain processes can read the superblock after the captain process writes it. H5FD.c - Error message correction. H5FDfphdf5.c - Removed MPI barrier call that wasn't needed. H5FPclient.c H5FPserver.c - Modified so that if a process requests data that isn't exactly aligned, we can return it if we have the block that contains the requested address. H5G.c H5Gent.c H5Gnode.c H5HL.c H5HLpkg.h H5HLprivate.h H5Oefl.c - Removed the H5HL_peek function since it was doing a (now unsafe) holding of the information in the cache. Replaced with protect and unprotect calls. H5TB.c - Error fix. The TB_dless function wasn't working properly. H5Gstab.c - Format change. H5err.txt H5Edefin.h H5Einit.h H5Epubgen.h H5Eterm.h - Added new error code. Platforms tested: Modi4 (paralle, Fortran) Sol (Fortran) Linux (C++, Fortran) Copper (Parallel, Fortran) Misc. update:
2003-09-19 03:27:27 +08:00
if (NULL == (heap = H5HL_protect(grp_ent->file, dxpl_id, stab_mesg.heap_addr)))
HGOTO_ERROR (H5E_SYM, H5E_NOTFOUND, FAIL, "unable to read protect link value")
clv = H5HL_offset_into(grp_ent->file, heap, obj_ent->cache.slink.lval_offset);
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
linkval = H5MM_xstrdup (clv);
[svn-r7489] Purpose: Update Description: A lot of modifications for the FPHDF5 stuff: H5AC.c H5ACprivate.h - Removed AC_find (it's replaced with AC_protect and AC_unprotect). Added flushing if it's an FPHDF5 driver and we're doing an AC_set or AC_unprotect with the dirty flag set. H5B.c - Split up the B_flush function into different functions since the one function was doing serialization which is better left as a separate entity. H5D.c - Removed some FPHDF5 code that was incorrect H5F.c - Split up the F_flush function so that it no longer allocates file space. Created new functions (F_init_superblock, F_read_superblock, and F_write_superblock) for greater modularity and so that the FPHDF5 non-captain processes can read the superblock after the captain process writes it. H5FD.c - Error message correction. H5FDfphdf5.c - Removed MPI barrier call that wasn't needed. H5FPclient.c H5FPserver.c - Modified so that if a process requests data that isn't exactly aligned, we can return it if we have the block that contains the requested address. H5G.c H5Gent.c H5Gnode.c H5HL.c H5HLpkg.h H5HLprivate.h H5Oefl.c - Removed the H5HL_peek function since it was doing a (now unsafe) holding of the information in the cache. Replaced with protect and unprotect calls. H5TB.c - Error fix. The TB_dless function wasn't working properly. H5Gstab.c - Format change. H5err.txt H5Edefin.h H5Einit.h H5Epubgen.h H5Eterm.h - Added new error code. Platforms tested: Modi4 (paralle, Fortran) Sol (Fortran) Linux (C++, Fortran) Copper (Parallel, Fortran) Misc. update:
2003-09-19 03:27:27 +08:00
assert(linkval);
if (H5HL_unprotect(grp_ent->file, dxpl_id, heap, stab_mesg.heap_addr) < 0)
HGOTO_ERROR (H5E_SYM, H5E_NOTFOUND, FAIL, "unable to read unprotect link value")
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Hold the entry's name (& old_name) to restore later */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
tmp_user_path_r=obj_ent->user_path_r;
obj_ent->user_path_r=NULL;
tmp_canon_path_r=obj_ent->canon_path_r;
obj_ent->canon_path_r=NULL;
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Free the names for the group entry */
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
H5G_free_ent_name(grp_ent);
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Clone the group entry, so we can track the names properly */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
H5G_ent_copy(&tmp_grp_ent,grp_ent,H5G_COPY_DEEP);
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/* Traverse the link */
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_namei (&tmp_grp_ent, linkval, NULL, grp_ent, obj_ent, H5G_TARGET_NORMAL, nlinks, H5G_NAMEI_TRAVERSE, NULL, dxpl_id))
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_NOTFOUND, FAIL, "unable to follow symbolic link");
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Free the entry's names, we will use the original name for the object */
H5G_free_ent_name(obj_ent);
/* Restore previous name for object */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
obj_ent->user_path_r = tmp_user_path_r;
tmp_user_path_r=NULL;
obj_ent->canon_path_r = tmp_canon_path_r;
tmp_canon_path_r=NULL;
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Error cleanup */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(tmp_user_path_r)
H5RS_decr(tmp_user_path_r);
if(tmp_canon_path_r)
H5RS_decr(tmp_canon_path_r);
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Release cloned copy of group entry */
H5G_free_ent_name(&tmp_grp_ent);
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
H5MM_xfree (linkval);
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
}
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
/*-------------------------------------------------------------------------
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Function: H5G_mkroot
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
* Purpose: Creates a root group in an empty file and opens it. If a
* root group is already open then this function immediately
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
* returns. If ENT is non-null then it's the symbol table
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
* entry for an existing group which will be opened as the root
* group. Otherwise a new root group is created and then
* opened.
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Return: Non-negative on success/Negative on failure
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Programmer: Robb Matzke
* matzke@llnl.gov
* Aug 11 1997
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
* Modifications:
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Pedro Vicente, <pvn@ncsa.uiuc.edu> 22 Aug 2002
* Added `id to name' support.
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
[svn-r61] ./src/H5.c Changed atexit() to HDatexit(). ./src/H5AC.c Changed qsort() to HDqsort(). ./src/H5B.c Changed memcmp() to HDmemcmp(). ./src/H5E.c ./src/H5M.c Added parentheses around indirect function calls so that checkposix doesn't complain and to advertise that it's an indirect function call. ./src/H5F.c ./src/H5G.c Added errors field to each function prologue to test the script mentioned above. ./src/H5H.c Changed fputc() to HDfputc(). ./src/H5MM.c ./src/H5MMprivate.h The argument for H5MM_xfree() is a pointer to a constant even though the argument to free() isn't. This reduces the number of warnings in other parts of the code but creates a new warning in this file. ./src/H5O.c Plugged a few memory leaks that happen during error handling. ./src/H5Oname.c ./src/debug.c Added `HD' to the beginning of some posix functions.
1997-09-02 23:38:26 +08:00
*-------------------------------------------------------------------------
*/
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
herr_t
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
H5G_mkroot (H5F_t *f, hid_t dxpl_id, H5G_entry_t *ent)
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
{
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
H5G_entry_t new_root; /*new root object */
H5O_stab_t stab; /*symbol table message */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_mkroot, FAIL);
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/* check args */
assert(f);
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (f->shared->root_grp)
HGOTO_DONE(SUCCEED);
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/*
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
* If there is no root object then create one. The root group always has
* a hard link count of one since it's pointed to by the boot block.
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
*/
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
if (!ent) {
ent = &new_root;
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
HDmemset(ent, 0, sizeof(H5G_entry_t));
[svn-r6654] Purpose: Code cleanup Description: Stop using hard-coded constant for the initial size of the root group's symbol table local heap and use macro. Also, changed existing (unused) macro for the initial heap size to be the same as the hard-coded constant. Platforms tested: FreeBSD 4.8 (sleipnir) w/C++ Linux 2.4 (burrwhite) w/FORTRAN Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/parallel & FORTRAN (h5committest not run due to my ongoing difficulties with C++ on burrwhite).
2003-04-14 12:49:26 +08:00
if (H5G_stab_create (f, dxpl_id, H5G_SIZE_HINT, ent/*out*/)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_CANTINIT, FAIL, "unable to create root group");
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (1 != H5O_link (ent, 1, dxpl_id))
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_LINK, FAIL, "internal error (wrong link count)");
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
} else {
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
/*
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
* Open the root object as a group.
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
*/
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (H5O_open (ent)<0)
HGOTO_ERROR (H5E_SYM, H5E_CANTOPENOBJ, FAIL, "unable to open root group");
[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
if (NULL==H5O_read (ent, H5O_STAB_ID, 0, &stab, dxpl_id)) {
[svn-r1087] Changes since 19990218 ---------------------- ./src/H5F.c ./src/H5private.h ./src/H5Ipublic.h ./src/H5O.c Fixed a rather nasty bug with file closing that caused the file boot block to be updated incorrectly, effectively truncating the file. The bug I fixed was triggered by: 1. Create a file, F 2. Open an object, X 3. Close file F 4. Reopen file F for read/write. 5. Create and close some objects 6. Close file F 7. Close library (exit). Step 3 pended the close because object X is still open, but the file ID was removed from the H5I_FILE ID group. Step 4 created a new file because it didn't see any matching file on the H5I_FILE ID group. Step 5 extends the file. Step 6 writes the new file boot block to disk. Step 7 closes object X and completes the close from step 3, writing the old boot block information to disk. The new behavior is that step 3 moves the file from the H5I_FILE group to the H5I_FILE_CLOSING group. Step 4 searches both groups and finds the file. Step 5 extends the file using the same H5F_file_t struct as step 3. Step 6 closes the H5F_t struct opened in step 3 but not the H5F_file_t struct shared by steps 1 and 3. Step 7 closes object X which closes the H5F_file_t from step 1, flushing the boot block which was shared by all steps. ./src/H5F.c Added some bulletproofing to file reference counting and removed comments which no longer apply. Added H5F_flush_all() and H5F_close_all() which apply to all files. ./src/H5A.c ./src/H5D.c ./src/H5F.c ./src/H5G.c ./src/H5I.c ./src/H5Iprivate.h ./src/H5R.c ./src/H5RA.c ./src/H5S.c ./src/H5T.c Added the new H5I_free_t data type to describe the function type to be passed as the `free_func' argument to H5I_init_group(). ./src/H5I.c Bulletproofed the object removal functions. Removed comments which no longer apply. Changed global variable names so they don't violate the naming scheme. Added H5I_debug() that prints the contents of an ID group. Removed H5I_inc_ref() because it isn't used. Reindented a couple of functions. ./src/H5.c ./src/H5G.c ./src/H5Ipublic.h Changed H5I_MAXID to H5I_NGROUPS to better relect the fact that it's the total number of valid ID groups. ./src/H5Shyper.c Changed hyperslab offset arrays to signed quantities to get rid of warnings on DEC cluster. ./src/H5Flow.c ./src/H5Fprivate.h Changed the objno argument of H5F_addr_pack() to be unsigned to get rid of warnings on DEC cluster.
1999-02-21 00:18:51 +08:00
H5O_close(ent);
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_NOTFOUND, FAIL, "root object is not a group");
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
}
[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
H5O_reset (H5O_STAB_ID, &stab);
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
}
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Create the path names for the root group's entry */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
ent->user_path_r=H5RS_create("/");
assert(ent->user_path_r);
ent->canon_path_r=H5RS_create("/");
assert(ent->canon_path_r);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
ent->user_path_hidden=0;
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
/*
* Create the group pointer. Also decrement the open object count so we
* don't count the root group as an open object. The root group will
* never be closed.
*/
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if (NULL==(f->shared->root_grp = H5FL_CALLOC (H5G_t)))
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed");
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
f->shared->root_grp->ent = *ent;
f->shared->root_grp->nref = 1;
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
assert (1==f->nopen_objs);
f->nopen_objs = 0;
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
}
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
/*-------------------------------------------------------------------------
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Function: H5G_create
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Purpose: Creates a new empty group with the specified name. The name
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
* is either an absolute name or is relative to LOC.
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
[svn-r61] ./src/H5.c Changed atexit() to HDatexit(). ./src/H5AC.c Changed qsort() to HDqsort(). ./src/H5B.c Changed memcmp() to HDmemcmp(). ./src/H5E.c ./src/H5M.c Added parentheses around indirect function calls so that checkposix doesn't complain and to advertise that it's an indirect function call. ./src/H5F.c ./src/H5G.c Added errors field to each function prologue to test the script mentioned above. ./src/H5H.c Changed fputc() to HDfputc(). ./src/H5MM.c ./src/H5MMprivate.h The argument for H5MM_xfree() is a pointer to a constant even though the argument to free() isn't. This reduces the number of warnings in other parts of the code but creates a new warning in this file. ./src/H5O.c Plugged a few memory leaks that happen during error handling. ./src/H5Oname.c ./src/debug.c Added `HD' to the beginning of some posix functions.
1997-09-02 23:38:26 +08:00
* Errors:
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Return: Success: A handle for the group. The group is opened
* and should eventually be close by calling
* H5G_close().
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Failure: NULL
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Programmer: Robb Matzke
* matzke@llnl.gov
* Aug 11 1997
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
* Modifications:
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Pedro Vicente, <pvn@ncsa.uiuc.edu> 18 Sep 2002
* Added `id to name' support.
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
*-------------------------------------------------------------------------
*/
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
static H5G_t *
H5G_create(H5G_entry_t *loc, const char *name, size_t size_hint, hid_t dxpl_id)
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
{
[svn-r7053] Purpose: Bug fix. Description: When a duplicate object was attempted to be created, the library would leak file memory and object references in the file, potentially causing an infinite loop when shutting the library down. Solution: Clean up after ourselves... :-) Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-06-19 04:41:28 +08:00
H5G_t *grp = NULL; /*new group */
[svn-r7393] Purpose: Code cleanup Description: De-linted more code Platforms tested: FreeBSD 4.8 (sleipnir) w/parallel too minor to require h5committest
2003-08-23 23:55:06 +08:00
H5F_t *file = NULL; /* File new group will be in */
[svn-r7053] Purpose: Bug fix. Description: When a duplicate object was attempted to be created, the library would leak file memory and object references in the file, potentially causing an infinite loop when shutting the library down. Solution: Clean up after ourselves... :-) Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-06-19 04:41:28 +08:00
unsigned stab_init=0; /* Flag to indicate that the symbol stable was created successfully */
H5G_t *ret_value; /* Return value */
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_create, NULL);
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/* check args */
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
assert(loc);
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
assert(name && *name);
/* create an open group */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if (NULL==(grp = H5FL_CALLOC(H5G_t)))
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_RESOURCE, H5E_NOSPACE, NULL, "memory allocation failed");
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/* What file is the group being added to? */
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (NULL==(file=H5G_insertion_file(loc, name, dxpl_id)))
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, NULL, "unable to locate insertion point");
/* Create the group entry */
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_stab_create(file, dxpl_id, size_hint, &(grp->ent)/*out*/) < 0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, NULL, "can't create grp");
[svn-r7053] Purpose: Bug fix. Description: When a duplicate object was attempted to be created, the library would leak file memory and object references in the file, potentially causing an infinite loop when shutting the library down. Solution: Clean up after ourselves... :-) Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-06-19 04:41:28 +08:00
stab_init=1; /* Indicate that the symbol table information is valid */
[svn-r328] Changes since 19980318 ---------------------- ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gpublic.h ./src/H5Gstab.c Implemented H5Giterate(). However, since most functions can't take a group ID in place of a file ID yet, there's not a whole lot that the operator can do besides print the name or something. ./test/H5O.c Fixed writing of four uninitialized bytes to the file as part of an object header. ./test/istore.c For some reason, `mpirun -np 1 istore' results in extra arguments on the command line that istore doesn't understand. I'm probably forgetting to call some MPI function in main(). Albert, Kim? So I commented out the `exit(1)' for the time being.
1998-03-24 22:53:57 +08:00
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/* insert child name into parent */
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if(H5G_insert(loc,name,&(grp->ent), dxpl_id)<0)
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINSERT, NULL, "can't insert group");
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
grp->nref = 1;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
/* Set return value */
ret_value=grp;
done:
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
if(ret_value==NULL) {
[svn-r7053] Purpose: Bug fix. Description: When a duplicate object was attempted to be created, the library would leak file memory and object references in the file, potentially causing an infinite loop when shutting the library down. Solution: Clean up after ourselves... :-) Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-06-19 04:41:28 +08:00
/* Check if we need to release the file-oriented symbol table info */
if(stab_init) {
if(H5O_close(&(grp->ent))<0)
HDONE_ERROR(H5E_SYM, H5E_CLOSEERROR, NULL, "unable to release object header");
if(H5O_delete(file, dxpl_id,grp->ent.header)<0)
HDONE_ERROR(H5E_SYM, H5E_CANTDELETE, NULL, "unable to delete object header");
} /* end if */
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
if(grp!=NULL)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
H5FL_FREE(H5G_t,grp);
} /* end if */
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
}
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_isa
*
* Purpose: Determines if an object has the requisite messages for being
* a group.
*
* Return: Success: TRUE if the required group messages are
* present; FALSE otherwise.
*
* Failure: FAIL if the existence of certain messages
* cannot be determined.
*
* Programmer: Robb Matzke
* Monday, November 2, 1998
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
static htri_t
H5G_isa(H5G_entry_t *ent, hid_t dxpl_id)
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
{
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
htri_t ret_value;
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_isa, FAIL);
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
assert(ent);
[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
if ((ret_value=H5O_exists(ent, H5O_STAB_ID, 0, dxpl_id))<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, FAIL, "unable to read object header");
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
}
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
/*-------------------------------------------------------------------------
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Function: H5G_open
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Purpose: Opens an existing group. The group should eventually be
* closed by calling H5G_close().
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Return: Success: Ptr to a new group.
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Failure: NULL
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Programmer: Robb Matzke
* Monday, January 5, 1998
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
* Modifications:
[svn-r1144] Added support for object references to groups.
1999-03-18 08:07:50 +08:00
* Modified to call H5G_open_oid - QAK - 3/17/99
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Pedro Vicente, <pvn@ncsa.uiuc.edu> 18 Sep 2002
* Added `id to name' support.
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
*
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*-------------------------------------------------------------------------
*/
[svn-r299] Changes since 19980224 ---------------------- ./html/Files.html Added a few details for some of the new H5Pset/get functions. ./src/H5F.c ./src/H5Fpublic.h Fixed automatic closing of files on exit(). Added public H5F_ACC_DEBUG. Using it to create or open a file turns on debugging for that file, which currently just prints cache statistics when the file is closed. ./src/H5G.c An error is returned if one tries to set the current working group to something other than a group. ./src/H5Gnode.c Fixed a symbol table bug. Under certain circumstances it was possible to enter a symbol in such a way that lookup of that symbol failed. A bug report was sent to hdf5dev. ./src/H5P.c ./src/H5Ppublic.h Added the H5Pget_...() functions for file drivers. The H5Pget_mpi() is a no-op that always fails.
1998-02-27 02:05:27 +08:00
H5G_t *
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
H5G_open(H5G_entry_t *loc, const char *name, hid_t dxpl_id)
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
{
[svn-r299] Changes since 19980224 ---------------------- ./html/Files.html Added a few details for some of the new H5Pset/get functions. ./src/H5F.c ./src/H5Fpublic.h Fixed automatic closing of files on exit(). Added public H5F_ACC_DEBUG. Using it to create or open a file turns on debugging for that file, which currently just prints cache statistics when the file is closed. ./src/H5G.c An error is returned if one tries to set the current working group to something other than a group. ./src/H5Gnode.c Fixed a symbol table bug. Under certain circumstances it was possible to enter a symbol in such a way that lookup of that symbol failed. A bug report was sent to hdf5dev. ./src/H5P.c ./src/H5Ppublic.h Added the H5Pget_...() functions for file drivers. The H5Pget_mpi() is a no-op that always fails.
1998-02-27 02:05:27 +08:00
H5G_t *grp = NULL;
H5G_t *ret_value = NULL;
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
H5G_entry_t ent; /* group symbol table entry */
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_open, NULL);
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/* Check args */
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
assert(loc);
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
assert(name && *name);
[svn-r299] Changes since 19980224 ---------------------- ./html/Files.html Added a few details for some of the new H5Pset/get functions. ./src/H5F.c ./src/H5Fpublic.h Fixed automatic closing of files on exit(). Added public H5F_ACC_DEBUG. Using it to create or open a file turns on debugging for that file, which currently just prints cache statistics when the file is closed. ./src/H5G.c An error is returned if one tries to set the current working group to something other than a group. ./src/H5Gnode.c Fixed a symbol table bug. Under certain circumstances it was possible to enter a symbol in such a way that lookup of that symbol failed. A bug report was sent to hdf5dev. ./src/H5P.c ./src/H5Ppublic.h Added the H5Pget_...() functions for file drivers. The H5Pget_mpi() is a no-op that always fails.
1998-02-27 02:05:27 +08:00
/* Open the object, making sure it's a group */
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_find(loc, name, NULL, &ent/*out*/, dxpl_id) < 0)
[svn-r1144] Added support for object references to groups.
1999-03-18 08:07:50 +08:00
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, NULL, "group not found");
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
[svn-r1144] Added support for object references to groups.
1999-03-18 08:07:50 +08:00
/* Open the group object */
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if ((grp=H5G_open_oid(&ent, dxpl_id)) ==NULL)
[svn-r1144] Added support for object references to groups.
1999-03-18 08:07:50 +08:00
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, NULL, "not found");
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
/* Set return value */
[svn-r1144] Added support for object references to groups.
1999-03-18 08:07:50 +08:00
ret_value = grp;
done:
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (!ret_value && grp)
[svn-r2476] Added free lists to track various data structures in memory, to reduce malloc abuse.
2000-08-17 04:13:02 +08:00
H5FL_FREE(H5G_t,grp);
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r1144] Added support for object references to groups.
1999-03-18 08:07:50 +08:00
}
/*-------------------------------------------------------------------------
* Function: H5G_open_oid
*
* Purpose: Opens an existing group. The group should eventually be
* closed by calling H5G_close().
*
* Return: Success: Ptr to a new group.
*
* Failure: NULL
*
* Programmer: Quincey Koziol
* Wednesday, March 17, 1999
*
* Modifications:
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Pedro Vicente, <pvn@ncsa.uiuc.edu> 22 Aug 2002
* Added a deep copy of the symbol table entry
[svn-r1144] Added support for object references to groups.
1999-03-18 08:07:50 +08:00
*
*-------------------------------------------------------------------------
*/
H5G_t *
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
H5G_open_oid(H5G_entry_t *ent, hid_t dxpl_id)
[svn-r1144] Added support for object references to groups.
1999-03-18 08:07:50 +08:00
{
H5G_t *grp = NULL;
H5G_t *ret_value = NULL;
H5O_stab_t mesg;
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_open_oid, NULL);
[svn-r1144] Added support for object references to groups.
1999-03-18 08:07:50 +08:00
/* Check args */
assert(ent);
/* Open the object, making sure it's a group */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if (NULL==(grp = H5FL_CALLOC(H5G_t)))
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_RESOURCE, H5E_NOSPACE, NULL, "memory allocation failed");
[svn-r1144] Added support for object references to groups.
1999-03-18 08:07:50 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Copy over (take ownership) of the group entry object */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
H5G_ent_copy(&(grp->ent),ent,H5G_COPY_SHALLOW);
[svn-r1144] Added support for object references to groups.
1999-03-18 08:07:50 +08:00
/* Grab the object header */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (H5O_open(&(grp->ent)) < 0)
[svn-r1144] Added support for object references to groups.
1999-03-18 08:07:50 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTOPENOBJ, NULL, "unable to open group");
[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
if (NULL==H5O_read (&(grp->ent), H5O_STAB_ID, 0, &mesg, dxpl_id)) {
[svn-r1144] Added support for object references to groups.
1999-03-18 08:07:50 +08:00
H5O_close(&(grp->ent));
HGOTO_ERROR (H5E_SYM, H5E_CANTOPENOBJ, NULL, "not a group");
[svn-r299] Changes since 19980224 ---------------------- ./html/Files.html Added a few details for some of the new H5Pset/get functions. ./src/H5F.c ./src/H5Fpublic.h Fixed automatic closing of files on exit(). Added public H5F_ACC_DEBUG. Using it to create or open a file turns on debugging for that file, which currently just prints cache statistics when the file is closed. ./src/H5G.c An error is returned if one tries to set the current working group to something other than a group. ./src/H5Gnode.c Fixed a symbol table bug. Under certain circumstances it was possible to enter a symbol in such a way that lookup of that symbol failed. A bug report was sent to hdf5dev. ./src/H5P.c ./src/H5Ppublic.h Added the H5Pget_...() functions for file drivers. The H5Pget_mpi() is a no-op that always fails.
1998-02-27 02:05:27 +08:00
}
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
grp->nref = 1;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
/* Set return value */
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
ret_value = grp;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
if (!ret_value && grp)
[svn-r2476] Added free lists to track various data structures in memory, to reduce malloc abuse.
2000-08-17 04:13:02 +08:00
H5FL_FREE(H5G_t,grp);
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r1184] Changes since 19990402 ---------------------- ./config/commence.in ./config/conclude.in ./test/Makefile.in ./tools/Makefile.in Fixed so private libraries are not installed publicly. The installation directories `bin', `include', and `lib' are created mode 755. ./src/H5.c ./src/H5A.c ./src/H5F.c ./src/H5Fcore.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fmpio.c ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5HG.c ./src/H5HL.c ./src/H5O.c ./src/H5Oattr.c ./src/H5Ocomp.c ./src/H5Ocont.c ./src/H5Odtype.c ./src/H5Oefl.c ./src/H5Ofill.c ./src/H5Olayout.c ./src/H5Omtime.c ./src/H5Oname.c ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5R.c ./src/H5Sall.c ./src/H5Shyper.c ./src/H5Smpio.c ./src/H5Snone.c ./src/H5Spoint.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Vprivate.h ./src/H5Z.c ./src/H5detect.c ./src/H5private.h ./test/chunk.c ./test/dsets.c ./test/dtypes.c ./test/h5test.c ./test/overhead.c ./test/ragged.c ./test/tattr.c ./tools/h5dump.c ./tools/h5findshd.c ./tools/h5ls.c Changed `__unused__' to `UNUSED' to fix a conflict with GNU header files. ./src/H5Tpkg.h ./test/h5test.h Removed __unused__ from forward function declarations. ./src/H5P.c Removed a comment about restrictions for the type conversion temporary buffers. Thanks to Quincey, the comment no longer applied. ./src/H5T.c Relaxed the H5Tpack() a little so it would pack compound data structures that had non-transient atomic members. ./tools/h5ls.c Added a `-g' (or `--group') flag that causes information to be printed about the specified group instead of the group's contents. (sort of like Unix's `ls -d'). The `-g' can be used in combination with `-r' to print information about the group and its contents.
1999-04-16 03:57:50 +08:00
}
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
#ifdef NOT_YET
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
/*-------------------------------------------------------------------------
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Function: H5G_reopen
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Purpose: Reopens a group by incrementing the open count.
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Return: Success: The GRP argument.
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Failure: NULL
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Programmer: Robb Matzke
* Monday, January 5, 1998
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
H5G_t *
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
H5G_reopen(H5G_t *grp)
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
{
[svn-r5867] Purpose: Code cleanup Description: Changed the last HRETURN* statements in the FUNC_ENTER macros into HGOTO* macros, which reduces the size of the library binary in certain configurations by another 10% Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel, IRIX64 6.5 (modi4) serial & parallel
2002-08-10 04:48:23 +08:00
H5G_t *ret_value; /* Return value */
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_reopen, NULL);
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
assert(grp);
assert(grp->nref > 0);
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
grp->nref++;
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
[svn-r5867] Purpose: Code cleanup Description: Changed the last HRETURN* statements in the FUNC_ENTER macros into HGOTO* macros, which reduces the size of the library binary in certain configurations by another 10% Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel, IRIX64 6.5 (modi4) serial & parallel
2002-08-10 04:48:23 +08:00
/* Set return value */
ret_value=grp;
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
}
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
#endif /* NOT_YET */
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
/*-------------------------------------------------------------------------
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Function: H5G_close
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Purpose: Closes the specified group.
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Return: Non-negative on success/Negative on failure
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Programmer: Robb Matzke
* Monday, January 5, 1998
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
herr_t
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
H5G_close(H5G_t *grp)
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
{
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_close, FAIL);
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/* Check args */
assert(grp);
assert(grp->nref > 0);
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
if (1 == grp->nref) {
[svn-r753] Changes since 19981008 ---------------------- ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./test/istore.c ./test/tstab.c Removed CWG functionality. ./tools/Makefile.in Added `_test' where I added `test' last week. Maybe now the snapshots will start up again...
1998-10-14 01:17:50 +08:00
assert (grp!=H5G_rootof(H5G_fileof(grp)));
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (H5O_close(&(grp->ent)) < 0)
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, FAIL, "unable to close");
[svn-r384] Changes since 19980430 ---------------------- ./src/H5G.c Fixed a read/write of free memory that my home-brew purify doesn't catch (yet :-)
1998-05-02 05:16:06 +08:00
grp->nref = 0;
[svn-r2476] Added free lists to track various data structures in memory, to reduce malloc abuse.
2000-08-17 04:13:02 +08:00
H5FL_FREE (H5G_t,grp);
[svn-r384] Changes since 19980430 ---------------------- ./src/H5G.c Fixed a read/write of free memory that my home-brew purify doesn't catch (yet :-)
1998-05-02 05:16:06 +08:00
} else {
--grp->nref;
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
}
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
}
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r100] Interim checkin for purify. Also contains H5G shadow stuff, changes to int64 encode/decode to fix purify on 32-bit systems, changes to H5O_modify, changes to H5D.c to use H5F_open/close(), etc. Documentation will be mailed shortly...
1997-09-20 00:36:59 +08:00
/*-------------------------------------------------------------------------
[svn-r753] Changes since 19981008 ---------------------- ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./test/istore.c ./test/tstab.c Removed CWG functionality. ./tools/Makefile.in Added `_test' where I added `test' last week. Maybe now the snapshots will start up again...
1998-10-14 01:17:50 +08:00
* Function: H5G_rootof
[svn-r100] Interim checkin for purify. Also contains H5G shadow stuff, changes to int64 encode/decode to fix purify on 32-bit systems, changes to H5O_modify, changes to H5D.c to use H5F_open/close(), etc. Documentation will be mailed shortly...
1997-09-20 00:36:59 +08:00
*
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
* Purpose: Return a pointer to the root group of the file. If the file
* is part of a virtual file then the root group of the virtual
* file is returned.
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
* Return: Success: Ptr to the root group of the file. Do not
* free the pointer -- it points directly into
* the file struct.
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
*
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
* Failure: NULL
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Programmer: Robb Matzke
[svn-r753] Changes since 19981008 ---------------------- ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./test/istore.c ./test/tstab.c Removed CWG functionality. ./tools/Makefile.in Added `_test' where I added `test' last week. Maybe now the snapshots will start up again...
1998-10-14 01:17:50 +08:00
* Tuesday, October 13, 1998
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
[svn-r6878] Purpose: Code cleanup Description: Limit the scope on more function prototypes/macros/typedefs. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest not necessary.
2003-05-16 03:22:33 +08:00
static H5G_t *
[svn-r753] Changes since 19981008 ---------------------- ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./test/istore.c ./test/tstab.c Removed CWG functionality. ./tools/Makefile.in Added `_test' where I added `test' last week. Maybe now the snapshots will start up again...
1998-10-14 01:17:50 +08:00
H5G_rootof(H5F_t *f)
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
{
[svn-r5867] Purpose: Code cleanup Description: Changed the last HRETURN* statements in the FUNC_ENTER macros into HGOTO* macros, which reduces the size of the library binary in certain configurations by another 10% Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel, IRIX64 6.5 (modi4) serial & parallel
2002-08-10 04:48:23 +08:00
H5G_t *ret_value; /* Return value */
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_rootof, NULL);
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
while (f->mtab.parent)
f = f->mtab.parent;
[svn-r5867] Purpose: Code cleanup Description: Changed the last HRETURN* statements in the FUNC_ENTER macros into HGOTO* macros, which reduces the size of the library binary in certain configurations by another 10% Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel, IRIX64 6.5 (modi4) serial & parallel
2002-08-10 04:48:23 +08:00
/* Set return value */
ret_value=f->shared->root_grp;
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
}
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
/*-------------------------------------------------------------------------
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Function: H5G_insert
[svn-r27] ./src/H5B.c ./src/H5Bprivate.h The B-tree K value comes from a combination of the B-tree subclass and the file. ./src/H5C.c ./src/H5F.c ./src/hdf5lims.h ./src/hdf5type.h Removed the B-tree size parameter and added an array of B-tree K values. Also added symbol table node K value. ./src/H5Eprivate.h ./src/H5Eproto.h Added H5E_LINK for errors involving link counts. ./src/H5G.c Inserting something into a directory with H5G_insert() increments the link count in the object header. The root object should always have a link count of at least 1. ./src/H5Gnode.c ./src/H5Gprivate.h The symbol table node K value comes from the file instead of being a constant. ./src/H5Olink.c Added an assert(), fixed a hard-link bug.
1997-08-13 23:36:47 +08:00
*
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
* Purpose: Inserts a symbol table entry into the group graph.
[svn-r61] ./src/H5.c Changed atexit() to HDatexit(). ./src/H5AC.c Changed qsort() to HDqsort(). ./src/H5B.c Changed memcmp() to HDmemcmp(). ./src/H5E.c ./src/H5M.c Added parentheses around indirect function calls so that checkposix doesn't complain and to advertise that it's an indirect function call. ./src/H5F.c ./src/H5G.c Added errors field to each function prologue to test the script mentioned above. ./src/H5H.c Changed fputc() to HDfputc(). ./src/H5MM.c ./src/H5MMprivate.h The argument for H5MM_xfree() is a pointer to a constant even though the argument to free() isn't. This reduces the number of warnings in other parts of the code but creates a new warning in this file. ./src/H5O.c Plugged a few memory leaks that happen during error handling. ./src/H5Oname.c ./src/debug.c Added `HD' to the beginning of some posix functions.
1997-09-02 23:38:26 +08:00
*
[svn-r129] Changes since 19970916 ---------------------- ./config/depend.in Fixed backslashes in sed script because the H5Gnode.c dependency info was disappearing. You'll have to rerun config.status to rebuild the Makefiles unless you use gnu make. ./config/conclude.in Also removes emacs backup files, TAGS, and svf backup files. ./config/linux Grouped gcc flags and added provisions for debugging vs. production. ./html/H5.format.html Updated messages 0x0008, 0x0009, and 0x000A. ./html/storage.html Documentation describing storage schemes. ./src/Makefile.in ./test/Makefile.in New source files. ./src/H5A.c ./src/H5Apublic.h ./src/H5C.c Changed VOIDP to void* in a couple places. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/H5H.c ./src/H5O.c Removed `const' from some variables because H5G_node_found() wanted to modify it's udata argument. Removing const there caused it to cascade to these other locations. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5B.c ./src/H5Gnode.c ./src/H5Gstab.c ./src/H5H.c ./src/H5O.c Added an extra argument to H5AC_find_f() and H5AC_protect(). This arg gets passed to the load() method. Also added an extra argument to the H5AC_find() macro. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra argument passed to the sizeof_rkey() method. ./src/H5Fprivate.c ./src/H5Fistore.c (new) Added indexed I/O operations. ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h Beginning to add H5G_open/close and related bug fixes. ./src/H5Oprivate.h ./src/H5Oistore.c (new) Added the H5O_ISTORE messsage (0x0008) for indexed storage of objects. ./src/H5private.h Added extra braces around both sides of the FUNC_ENTER() and FUNC_LEAVE() macros so FUNC_ENTER() can appear before declarations or after executable statements the second case is used by H5G_namei() to initialize output arguments to sane values before FUNC_ENTER() might return failure. int f () { int decl1; printf ("This happens before FUNC_ENTER()\n"); FUNC_ENTER (...); int another_declaration; ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c Extra arguments for key encoding and decoding. ./src/H5E.c ./src/H5Epublic.h ./src/H5Fistore.c ./src/H5Oistore.c ./src/H5Oprivate.h Indexed, chunked, sparse storage (not ready for general consumption yet). ./src/H5V.c (new) ./src/H5Vprivate.h (new) ./test/hyperslab.c (new) Vector, array, and hyperslab functions. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5Gnode.c ./src/H5V.c ./src/H5Vprivate.h ./test/hyperslab.c Added functionality for indexed storage. ./src/H5F.c Fixed problems with seek optimizing. Recommend we disable it until we can implement it in the file/address class since all of HDF5 must be aware of it. ./src/H5O.c Fixed comeent speling erorr :-) ./MANIFEST Added new files. ./config/conclude.in Added the word `Testing' to the test cases. So if a test program is called hyperslab then the make output will contain the line `Testing hyperslab'. ./config/linux The default file I/O library is Posix section 2 on my linux machine so I can do some I/O performance testing. ./src/H5C.c ./src/H5Cprivate.h ./src/H5Cpublic.h Added ability to set size of indexed-storage B-tree. ./src/H5D.c ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c ./src/H5Gstab.c ./test/stab.c Changed `directory' to `group' in numerous places. ./src/H5private.h The FILELIB constant can be set on the compile command-line. ./src/istore.c NEW Tests for indexed storage.
1997-10-21 07:14:35 +08:00
* Errors:
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Return: Non-negative on success/Negative on failure
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Programmer: Robb Matzke
* Friday, September 19, 1997
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
*
* Modifications:
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Pedro Vicente, <pvn@ncsa.uiuc.edu> 18 Sep 2002
* Added `id to name' support.
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
*
[svn-r61] ./src/H5.c Changed atexit() to HDatexit(). ./src/H5AC.c Changed qsort() to HDqsort(). ./src/H5B.c Changed memcmp() to HDmemcmp(). ./src/H5E.c ./src/H5M.c Added parentheses around indirect function calls so that checkposix doesn't complain and to advertise that it's an indirect function call. ./src/H5F.c ./src/H5G.c Added errors field to each function prologue to test the script mentioned above. ./src/H5H.c Changed fputc() to HDfputc(). ./src/H5MM.c ./src/H5MMprivate.h The argument for H5MM_xfree() is a pointer to a constant even though the argument to free() isn't. This reduces the number of warnings in other parts of the code but creates a new warning in this file. ./src/H5O.c Plugged a few memory leaks that happen during error handling. ./src/H5Oname.c ./src/debug.c Added `HD' to the beginning of some posix functions.
1997-09-02 23:38:26 +08:00
*-------------------------------------------------------------------------
*/
[svn-r150] Changes since 19971219 ---------------------- ./src/H5private.h Changed HDF5_MAJOR_VERSION to 5 ./src/H5Aprivate.h ./src/H5Apublic.h Added group atoms. ./src/H5D.c ./src/H5P.c ./test/istore.c ./test/tohdr.c ./test/tstab.c Updated for symbol table interface changes. ./src/H5F.c ./src/H5Fprivate.h Simpler handling of special case files: empty files and files with a single object. ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5Gshad.c (DELETED) ./src/H5Gstab.c ./src/Makefile.in Removed shadows, simplifying code. Symbol table entries are allowed to cache only constant meta data. Fixed naming. ./src/H5O.c ./src/H5Oprivate.h Access to object headers is always done through a symbol table entry instead of a file address. Added stubs for opening and closing object headers to be used when deletion is implemented.
1998-01-06 11:07:15 +08:00
herr_t
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
H5G_insert(H5G_entry_t *loc, const char *name, H5G_entry_t *ent, hid_t dxpl_id)
[svn-r25] ./src/H5AC.c We sort the cache before a complete flush because it might be more efficient to write things back to disk in order of increasing address. If you want the old way then undef the SORT_BY_ADDR constant at the top of H5AC.c I haven't determined which systems and I/O libraries this helps or hurts. (This is currently off because of a bug I need to track down that causes qsort() to run for a really long time). ./src/H5B.c Fixed a couple more bugs. ./src/H5Eprivate.h ./src/H5Eproto.h Added major H5E_DIRECTORY and minor H5E_EXISTS, H5E_COMPLEN. ./src/H5G.c Added directory-aware functions. The heap and B-tree are created when a directory is created instead of when the first symbol is added. This simplifies symbol table entry caching for the directory since the cached value never changes now. ./src/H5Gnode.c ./src/H5Gprivate.h Fine tuned the B-tree K values for symbol tables assuming an average number of symbols is about 100 per directory. The tuning minimizes storage space. Fixed a return value in H5G_node_cmp(). ./src/H5H.c ./src/H5Hprivate.h Moved some macros the the header file. ./src/H5O.c ./src/H5Ocont.c ./src/H5Onull.c ./src/H5Ostab.c Changed the arguments for the decode method for messages. The second argument is the raw message size. Added a class variable for native message size. Added H5O_reset() to free memory used internally by a message. ./src/H5Oname.c NEW ./src/H5Oprivate.h ./src/Makefile The object name message. ./src/hdf5port.h Added defn for HDstrdup()
1997-08-13 06:44:46 +08:00
{
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_insert, FAIL);
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/* Check args. */
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
assert (loc);
assert (name && *name);
assert (ent);
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/*
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
* Lookup and insert the name -- it shouldn't exist yet.
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
*/
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_namei(loc, name, NULL, NULL, NULL, H5G_TARGET_NORMAL, NULL, H5G_NAMEI_INSERT, ent, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_EXISTS, FAIL, "already exists");
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/*
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
* Insert the object into a symbol table.
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
*/
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5O_link(ent, 1, dxpl_id) < 0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_LINK, FAIL, "unable to increment hard link count");
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r2] Oops...
1997-07-31 05:17:56 +08:00
}
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r2] Oops...
1997-07-31 05:17:56 +08:00
/*-------------------------------------------------------------------------
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Function: H5G_find
*
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
* Purpose: Finds an object with the specified NAME at location LOC. On
* successful return, GRP_ENT (if non-null) will be initialized
* with the symbol table information for the group in which the
* object appears (it will have an undefined object header
* address if the object is the root object) and OBJ_ENT will be
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* initialized with the symbol table entry for the object
* (OBJ_ENT is optional when the caller is interested only in
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
* the existence of the object).
[svn-r2] Oops...
1997-07-31 05:17:56 +08:00
*
[svn-r61] ./src/H5.c Changed atexit() to HDatexit(). ./src/H5AC.c Changed qsort() to HDqsort(). ./src/H5B.c Changed memcmp() to HDmemcmp(). ./src/H5E.c ./src/H5M.c Added parentheses around indirect function calls so that checkposix doesn't complain and to advertise that it's an indirect function call. ./src/H5F.c ./src/H5G.c Added errors field to each function prologue to test the script mentioned above. ./src/H5H.c Changed fputc() to HDfputc(). ./src/H5MM.c ./src/H5MMprivate.h The argument for H5MM_xfree() is a pointer to a constant even though the argument to free() isn't. This reduces the number of warnings in other parts of the code but creates a new warning in this file. ./src/H5O.c Plugged a few memory leaks that happen during error handling. ./src/H5Oname.c ./src/debug.c Added `HD' to the beginning of some posix functions.
1997-09-02 23:38:26 +08:00
* Errors:
*
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
* Return: Success: Non-negative, see above for values of GRP_ENT
* and OBJ_ENT.
[svn-r2] Oops...
1997-07-31 05:17:56 +08:00
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Failure: Negative
[svn-r2] Oops...
1997-07-31 05:17:56 +08:00
*
[svn-r178] Changes since 19980127 ---------------------- ./Makefile.in ./config/commence.in With GNU make you can now use `-j' and `-l' options and things get built correctly. I can do a `make -j -l6 test' from a clean hdf5 source tree (after configure) in 45 seconds (8 to build dependencies, 26 to compile everything, and 11 to run the tests). ./src/H5Gnode.c Removed a comment that no longer applies. ./src/H5P.c ./src/H5Pprivate.h ./src/H5Ppublic.h Changed H5Pselect_hyperslab() to H5Pset_hyperslab() and added H5Pget_hyperslab() and H5P_get_hyperslab(). Replaced a couple short memset() calls with a for loop. Removed `if (foo!=NULL)' from around H5MM_xfree() calls. Clear hslab_def when the hyperslab disappears. ./src/H5Tpublic.h Removed trailing enum comma.
1998-01-28 13:47:19 +08:00
* Programmer: Robb Matzke
* matzke@llnl.gov
* Aug 12 1997
[svn-r2] Oops...
1997-07-31 05:17:56 +08:00
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
herr_t
[svn-r633] Changes since 19980828 ---------------------- ./RELEASE Updated with important changes I made since the second beta. ./src/H5A.c ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5R.c ./src/H5Rprivate.h ./src/H5T.c ./src/H5Tprivate.h Any API function that used to take an `hid_t loc_id' followed by a `const char *name' can now take any type of object for the loc_id as long as the object is somehow associated with a file. Internally, H5G_loc() was modified to return an H5G_entry_t* instead of an H5G_t* so it's more general. Among other things, this allows one to retrieve information about an object like a named type or dataset without knowing the name of the type or dataset: int get_nlinks (hid_t obj) { H5G_stat_t sb; if (H5Gstat(obj, ".", TRUE, &sb)<0) return -1; return sb.nlink; } ./test/gheap.c ./test/istore.c These files needed a couple of changes because they call some of the internal functions whose H5G_t arguments changed to H5G_entry_t. ./src/H5A.c Got rid of all the switch statements for getting symbol table entries for varous objects and replaced them with a call to H5G_loc() allowing attributes to automatically apply to any type of object that belongs to a file. ./test/Makefile.in Moved the ragged array tests from the normal list of tests to the `make timings' target. ./test/ragged.c Added rewrite tests -- rewrite the rows of a dataset changing the number of rows and the length of each row. ./test/mtime.c Added a test that checks that H5Gstat() can be called with a dataset as the first argument. ./src/H5S.c Added #ifdef HAVE_PARALLEL around code to check for the HDF5_MPI_OPT_TYPES environment variable because the global variable that gets set is #ifdef'd. ./bin/release bzip2 uses .bz2 as the file extension.
1998-08-31 21:46:47 +08:00
H5G_find(H5G_entry_t *loc, const char *name,
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
H5G_entry_t *grp_ent/*out*/, H5G_entry_t *obj_ent/*out*/, hid_t dxpl_id)
[svn-r2] Oops...
1997-07-31 05:17:56 +08:00
{
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_find, FAIL);
[svn-r15] ./src/H5AC.c ./src/H5ACprivate.h ./src/H5ACproto.h ./src/H5Bproto.h ./src/H5Gproto.h ./src/H5Hproto.h ./src/H5MFprivate.h ./src/H5MFproto.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5MMproto.h Changed my e-mail address. ./src/H5O.c NEW ./src/H5Onull.c NEW ./src/H5Ocont. NEW ./src/H5Ostab.c NEW New functions for dealing with object headers. The H5O.c is the generic stuff, and each particular message has a source file for the specific stuff. Use ./src/H5Ostab.c as a model for implementing other messages. ./src/Makefile ./src/test/Makefile Added new files ./src/debug.c Added debugging calls for object headers. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c ./src/H5MF.c Changed my e-mail address. Improved error handling. ./src/H5Eprivate.h ./src/H5Eproto.h Added more error symbols ./src/H5F.c Changed my e-mail address. Used macros for sizeof offsets and lengths. Added the interface initialization function to H5F_block_read() and H5F_block_write(). Updated H5F_debug() ./src/H5Fprivate.h Got rid of H5F_symbol_table_size(). Use H5G_SIZEOF_ENTRY() instead. Reformatted H5F_decode_offset() for readability. ./src/H5G.c Changed my e-mail address. Improved error handling. Replaced not_implemented_yet__*() with real functions from H5O. ./src/H5Gprivate.h Changed `symtab' to `stab' to be consistent with other stuff. ./src/H5H.c ./src/H5Hprivate.h Changed my e-mail address. Improved error handling. Added an extra argument to H5H_new() to indicate whether you want a global heap or a local heap. ./src/hdf5gen.h Added NELMTS() Fixed FUNC_ENTER() Rewrote FUNC_LEAVE() in terms of HRETURN(). ./src/hdf5meta.h Added `const' to the decode macros. ./src/hdf5pabl.h Added PABLO_SAVE() ./test/testhdf5.c ./test/testhdf5.h Added calls for object header testing. ./test/theap.c Turned off some output.
1997-08-08 03:23:00 +08:00
[svn-r157] Reformatted code with indent...
1998-01-17 06:23:43 +08:00
/* check args */
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
assert (loc);
assert (name && *name);
[svn-r15] ./src/H5AC.c ./src/H5ACprivate.h ./src/H5ACproto.h ./src/H5Bproto.h ./src/H5Gproto.h ./src/H5Hproto.h ./src/H5MFprivate.h ./src/H5MFproto.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5MMproto.h Changed my e-mail address. ./src/H5O.c NEW ./src/H5Onull.c NEW ./src/H5Ocont. NEW ./src/H5Ostab.c NEW New functions for dealing with object headers. The H5O.c is the generic stuff, and each particular message has a source file for the specific stuff. Use ./src/H5Ostab.c as a model for implementing other messages. ./src/Makefile ./src/test/Makefile Added new files ./src/debug.c Added debugging calls for object headers. ./src/H5B.c ./src/H5Bprivate.h ./src/H5Gnode.c ./src/H5MF.c Changed my e-mail address. Improved error handling. ./src/H5Eprivate.h ./src/H5Eproto.h Added more error symbols ./src/H5F.c Changed my e-mail address. Used macros for sizeof offsets and lengths. Added the interface initialization function to H5F_block_read() and H5F_block_write(). Updated H5F_debug() ./src/H5Fprivate.h Got rid of H5F_symbol_table_size(). Use H5G_SIZEOF_ENTRY() instead. Reformatted H5F_decode_offset() for readability. ./src/H5G.c Changed my e-mail address. Improved error handling. Replaced not_implemented_yet__*() with real functions from H5O. ./src/H5Gprivate.h Changed `symtab' to `stab' to be consistent with other stuff. ./src/H5H.c ./src/H5Hprivate.h Changed my e-mail address. Improved error handling. Added an extra argument to H5H_new() to indicate whether you want a global heap or a local heap. ./src/hdf5gen.h Added NELMTS() Fixed FUNC_ENTER() Rewrote FUNC_LEAVE() in terms of HRETURN(). ./src/hdf5meta.h Added `const' to the decode macros. ./src/hdf5pabl.h Added PABLO_SAVE() ./test/testhdf5.c ./test/testhdf5.h Added calls for object header testing. ./test/theap.c Turned off some output.
1997-08-08 03:23:00 +08:00
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_namei(loc, name, NULL, grp_ent, obj_ent, H5G_TARGET_NORMAL, NULL, H5G_NAMEI_TRAVERSE, NULL, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, FAIL, "object not found");
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r2] Oops...
1997-07-31 05:17:56 +08:00
}
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_entof
*
* Purpose: Returns a pointer to the entry for a group.
*
* Return: Success: Ptr to group entry
*
* Failure: NULL
*
* Programmer: Robb Matzke
* Tuesday, March 24, 1998
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
H5G_entry_t *
H5G_entof (H5G_t *grp)
{
[svn-r5467] Purpose: Code cleanup. Description: Took Robb's recent ideas for improving the FUNC_ENTER/FUNC_LEAVE macros equivalents in the SAF library and adapted them to our library. I added an additional macro which is equivalent to FUNC_ENTER: FUNC_ENTER_NOINIT - Has the API tracing code, etc. from FUNC_ENTER but none of the library or interface initialization code. This is to be used _only_ for static functions and those which explicitly cannot have the library or interface initialization code enabled (like the API termination routines, etc.). This allowed many more of the functions in the library [but not all yet :-(] to be wrapped with FUNC_ENTER[_NOINIT]/FUNC_LEAVE pairs. It also reduced the size of the library and executables (by cutting out a bunch of code which was never executed), I'll e-mail the exact results when I've finished editing it. Platforms tested: IRIX64 6.5 (modi4)
2002-05-29 02:17:12 +08:00
/* Use FUNC_ENTER_NOINIT here to avoid performance issues */
FUNC_ENTER_NOINIT(H5G_entof);
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(grp ? &(grp->ent) : NULL);
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
}
/*-------------------------------------------------------------------------
* Function: H5G_fileof
*
* Purpose: Returns the file to which the specified group belongs.
*
* Return: Success: File pointer.
*
* Failure: NULL
*
* Programmer: Robb Matzke
* Tuesday, March 24, 1998
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
H5F_t *
H5G_fileof (H5G_t *grp)
{
[svn-r5467] Purpose: Code cleanup. Description: Took Robb's recent ideas for improving the FUNC_ENTER/FUNC_LEAVE macros equivalents in the SAF library and adapted them to our library. I added an additional macro which is equivalent to FUNC_ENTER: FUNC_ENTER_NOINIT - Has the API tracing code, etc. from FUNC_ENTER but none of the library or interface initialization code. This is to be used _only_ for static functions and those which explicitly cannot have the library or interface initialization code enabled (like the API termination routines, etc.). This allowed many more of the functions in the library [but not all yet :-(] to be wrapped with FUNC_ENTER[_NOINIT]/FUNC_LEAVE pairs. It also reduced the size of the library and executables (by cutting out a bunch of code which was never executed), I'll e-mail the exact results when I've finished editing it. Platforms tested: IRIX64 6.5 (modi4)
2002-05-29 02:17:12 +08:00
/* Use FUNC_ENTER_NOINIT here to avoid performance issues */
FUNC_ENTER_NOINIT(H5G_fileof);
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
assert (grp);
[svn-r5467] Purpose: Code cleanup. Description: Took Robb's recent ideas for improving the FUNC_ENTER/FUNC_LEAVE macros equivalents in the SAF library and adapted them to our library. I added an additional macro which is equivalent to FUNC_ENTER: FUNC_ENTER_NOINIT - Has the API tracing code, etc. from FUNC_ENTER but none of the library or interface initialization code. This is to be used _only_ for static functions and those which explicitly cannot have the library or interface initialization code enabled (like the API termination routines, etc.). This allowed many more of the functions in the library [but not all yet :-(] to be wrapped with FUNC_ENTER[_NOINIT]/FUNC_LEAVE pairs. It also reduced the size of the library and executables (by cutting out a bunch of code which was never executed), I'll e-mail the exact results when I've finished editing it. Platforms tested: IRIX64 6.5 (modi4)
2002-05-29 02:17:12 +08:00
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(grp->ent.file);
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
}
/*-------------------------------------------------------------------------
* Function: H5G_loc
*
[svn-r633] Changes since 19980828 ---------------------- ./RELEASE Updated with important changes I made since the second beta. ./src/H5A.c ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5R.c ./src/H5Rprivate.h ./src/H5T.c ./src/H5Tprivate.h Any API function that used to take an `hid_t loc_id' followed by a `const char *name' can now take any type of object for the loc_id as long as the object is somehow associated with a file. Internally, H5G_loc() was modified to return an H5G_entry_t* instead of an H5G_t* so it's more general. Among other things, this allows one to retrieve information about an object like a named type or dataset without knowing the name of the type or dataset: int get_nlinks (hid_t obj) { H5G_stat_t sb; if (H5Gstat(obj, ".", TRUE, &sb)<0) return -1; return sb.nlink; } ./test/gheap.c ./test/istore.c These files needed a couple of changes because they call some of the internal functions whose H5G_t arguments changed to H5G_entry_t. ./src/H5A.c Got rid of all the switch statements for getting symbol table entries for varous objects and replaced them with a call to H5G_loc() allowing attributes to automatically apply to any type of object that belongs to a file. ./test/Makefile.in Moved the ragged array tests from the normal list of tests to the `make timings' target. ./test/ragged.c Added rewrite tests -- rewrite the rows of a dataset changing the number of rows and the length of each row. ./test/mtime.c Added a test that checks that H5Gstat() can be called with a dataset as the first argument. ./src/H5S.c Added #ifdef HAVE_PARALLEL around code to check for the HDF5_MPI_OPT_TYPES environment variable because the global variable that gets set is #ifdef'd. ./bin/release bzip2 uses .bz2 as the file extension.
1998-08-31 21:46:47 +08:00
* Purpose: Given an object ID return a symbol table entry for the
* object.
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
*
* Return: Success: Group pointer.
*
* Failure: NULL
*
* Programmer: Robb Matzke
* Tuesday, March 24, 1998
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
[svn-r633] Changes since 19980828 ---------------------- ./RELEASE Updated with important changes I made since the second beta. ./src/H5A.c ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5R.c ./src/H5Rprivate.h ./src/H5T.c ./src/H5Tprivate.h Any API function that used to take an `hid_t loc_id' followed by a `const char *name' can now take any type of object for the loc_id as long as the object is somehow associated with a file. Internally, H5G_loc() was modified to return an H5G_entry_t* instead of an H5G_t* so it's more general. Among other things, this allows one to retrieve information about an object like a named type or dataset without knowing the name of the type or dataset: int get_nlinks (hid_t obj) { H5G_stat_t sb; if (H5Gstat(obj, ".", TRUE, &sb)<0) return -1; return sb.nlink; } ./test/gheap.c ./test/istore.c These files needed a couple of changes because they call some of the internal functions whose H5G_t arguments changed to H5G_entry_t. ./src/H5A.c Got rid of all the switch statements for getting symbol table entries for varous objects and replaced them with a call to H5G_loc() allowing attributes to automatically apply to any type of object that belongs to a file. ./test/Makefile.in Moved the ragged array tests from the normal list of tests to the `make timings' target. ./test/ragged.c Added rewrite tests -- rewrite the rows of a dataset changing the number of rows and the length of each row. ./test/mtime.c Added a test that checks that H5Gstat() can be called with a dataset as the first argument. ./src/H5S.c Added #ifdef HAVE_PARALLEL around code to check for the HDF5_MPI_OPT_TYPES environment variable because the global variable that gets set is #ifdef'd. ./bin/release bzip2 uses .bz2 as the file extension.
1998-08-31 21:46:47 +08:00
H5G_entry_t *
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
H5G_loc (hid_t loc_id)
{
H5F_t *f;
[svn-r5867] Purpose: Code cleanup Description: Changed the last HRETURN* statements in the FUNC_ENTER macros into HGOTO* macros, which reduces the size of the library binary in certain configurations by another 10% Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel, IRIX64 6.5 (modi4) serial & parallel
2002-08-10 04:48:23 +08:00
H5G_entry_t *ret_value=NULL;
[svn-r633] Changes since 19980828 ---------------------- ./RELEASE Updated with important changes I made since the second beta. ./src/H5A.c ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5R.c ./src/H5Rprivate.h ./src/H5T.c ./src/H5Tprivate.h Any API function that used to take an `hid_t loc_id' followed by a `const char *name' can now take any type of object for the loc_id as long as the object is somehow associated with a file. Internally, H5G_loc() was modified to return an H5G_entry_t* instead of an H5G_t* so it's more general. Among other things, this allows one to retrieve information about an object like a named type or dataset without knowing the name of the type or dataset: int get_nlinks (hid_t obj) { H5G_stat_t sb; if (H5Gstat(obj, ".", TRUE, &sb)<0) return -1; return sb.nlink; } ./test/gheap.c ./test/istore.c These files needed a couple of changes because they call some of the internal functions whose H5G_t arguments changed to H5G_entry_t. ./src/H5A.c Got rid of all the switch statements for getting symbol table entries for varous objects and replaced them with a call to H5G_loc() allowing attributes to automatically apply to any type of object that belongs to a file. ./test/Makefile.in Moved the ragged array tests from the normal list of tests to the `make timings' target. ./test/ragged.c Added rewrite tests -- rewrite the rows of a dataset changing the number of rows and the length of each row. ./test/mtime.c Added a test that checks that H5Gstat() can be called with a dataset as the first argument. ./src/H5S.c Added #ifdef HAVE_PARALLEL around code to check for the HDF5_MPI_OPT_TYPES environment variable because the global variable that gets set is #ifdef'd. ./bin/release bzip2 uses .bz2 as the file extension.
1998-08-31 21:46:47 +08:00
H5G_t *group=NULL;
H5T_t *dt=NULL;
H5D_t *dset=NULL;
H5A_t *attr=NULL;
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_loc, NULL);
[svn-r633] Changes since 19980828 ---------------------- ./RELEASE Updated with important changes I made since the second beta. ./src/H5A.c ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5R.c ./src/H5Rprivate.h ./src/H5T.c ./src/H5Tprivate.h Any API function that used to take an `hid_t loc_id' followed by a `const char *name' can now take any type of object for the loc_id as long as the object is somehow associated with a file. Internally, H5G_loc() was modified to return an H5G_entry_t* instead of an H5G_t* so it's more general. Among other things, this allows one to retrieve information about an object like a named type or dataset without knowing the name of the type or dataset: int get_nlinks (hid_t obj) { H5G_stat_t sb; if (H5Gstat(obj, ".", TRUE, &sb)<0) return -1; return sb.nlink; } ./test/gheap.c ./test/istore.c These files needed a couple of changes because they call some of the internal functions whose H5G_t arguments changed to H5G_entry_t. ./src/H5A.c Got rid of all the switch statements for getting symbol table entries for varous objects and replaced them with a call to H5G_loc() allowing attributes to automatically apply to any type of object that belongs to a file. ./test/Makefile.in Moved the ragged array tests from the normal list of tests to the `make timings' target. ./test/ragged.c Added rewrite tests -- rewrite the rows of a dataset changing the number of rows and the length of each row. ./test/mtime.c Added a test that checks that H5Gstat() can be called with a dataset as the first argument. ./src/H5S.c Added #ifdef HAVE_PARALLEL around code to check for the HDF5_MPI_OPT_TYPES environment variable because the global variable that gets set is #ifdef'd. ./bin/release bzip2 uses .bz2 as the file extension.
1998-08-31 21:46:47 +08:00
[svn-r737] Changed H5I groups to have a H5I_ prefix before them, to better align with the library.
1998-10-06 07:41:30 +08:00
switch (H5I_get_type(loc_id)) {
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
case H5I_FILE:
if (NULL==(f=H5I_object (loc_id)))
HGOTO_ERROR (H5E_ARGS, H5E_BADVALUE, NULL, "invalid file ID");
if (NULL==(ret_value=H5G_entof(H5G_rootof(f))))
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, NULL, "unable to get symbol table entry for root group");
break;
case H5I_GENPROP_CLS:
case H5I_GENPROP_LST:
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, NULL, "unable to get symbol table entry of property list");
[svn-r7265] *** empty log message ***
2003-07-26 10:55:47 +08:00
case H5I_ERROR_CLASS:
case H5I_ERROR_MSG:
case H5I_ERROR_STACK:
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, NULL, "unable to get symbol table entry of error class, message or stack");
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
case H5I_GROUP:
if (NULL==(group=H5I_object (loc_id)))
HGOTO_ERROR (H5E_ARGS, H5E_BADVALUE, NULL, "invalid group ID");
if (NULL==(ret_value=H5G_entof(group)))
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, NULL, "unable to get symbol table entry of group");
break;
case H5I_DATATYPE:
if (NULL==(dt=H5I_object(loc_id)))
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, NULL, "invalid type ID");
if (NULL==(ret_value=H5T_entof(dt)))
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, NULL, "unable to get symbol table entry of datatype");
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
break;
case H5I_DATASPACE:
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, NULL, "unable to get symbol table entry of dataspace");
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
case H5I_DATASET:
if (NULL==(dset=H5I_object(loc_id)))
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, NULL, "invalid data ID");
if (NULL==(ret_value=H5D_entof(dset)))
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, NULL, "unable to get symbol table entry of dataset");
break;
case H5I_ATTR:
if (NULL==(attr=H5I_object(loc_id)))
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, NULL, "invalid attribute ID");
if (NULL==(ret_value=H5A_entof(attr)))
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, NULL, "unable to get symbol table entry of attribute");
break;
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
case H5I_REFERENCE:
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, NULL, "unable to get symbol table entry of reference");
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
case H5I_NGROUPS:
case H5I_BADID:
case H5I_FILE_CLOSING:
case H5I_VFL:
HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, NULL, "invalid object ID");
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
}
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r329] Changes since 19980324 ---------------------- ./src/H5D.c Zero element requests for H5Dread() and H5Dwrite() succeed. ./src/H5F.c ./src/H5Fprivate.h All files will have a root group. This greatly simplifies the library at the expense of ~1k extra bytes in files that have only one object. ./src/H5D.c ./src/H5Dprivate.h ./html/Groups.html Functions that used to take a file ID and an object name can now take a group ID instead of a file ID. This doesn't change the API, only it's documentation. ./src/H5G.c ./src/H5Gprivate.h ./src/H5O.c ./src/H5Oprivate.h Removed extra file arguments from some internal functions since the file pointer can be found from the group pointer or a symbol table entry (ent->file or H5G_fileof(group)). Besides, when we eventually implement mounting one file on another, H5G_namei() might return a different file than what you gave it, and that file is part of the returned symbol table entry. ./src/H5G.c Fixed bug with `.' appearing in a name. It used to hang the library. ./src/Makefile.in ./src/h5ls.c [NEW] ./MANIFEST Added `h5ls' a simple program that takes a file name and a directory and lists the contents of that directory using H5Giterate(). ./test/istore.c Changed an argument to H5G_create(). ./test/tstab.c Removed test_1 which was testing that files with a single object don't have a root group. This no longer applies.
1998-03-25 07:18:34 +08:00
}
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_link
*
* Purpose: Creates a link from NEW_NAME to CUR_NAME. See H5Glink() for
* full documentation.
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Return: Non-negative on success/Negative on failure
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
*
* Programmer: Robb Matzke
* Monday, April 6, 1998
*
* Modifications:
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Pedro Vicente, <pvn@ncsa.uiuc.edu> 18 Sep 2002
* Added `id to name' support.
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
*
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
*-------------------------------------------------------------------------
*/
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
static herr_t
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
H5G_link (H5G_entry_t *cur_loc, const char *cur_name, H5G_entry_t *new_loc,
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
const char *new_name, H5G_link_t type, unsigned namei_flags, hid_t dxpl_id)
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
{
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
H5G_entry_t cur_obj; /*entry for the link tail */
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
unsigned cur_obj_init=0; /* Flag to indicate that the current object is initialized */
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
H5G_entry_t grp_ent; /*ent for grp containing link hd*/
H5O_stab_t stab_mesg; /*symbol table message */
const char *rest = NULL; /*last component of new name */
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
char *norm_cur_name = NULL; /* Pointer to normalized current name */
char *norm_new_name = NULL; /* Pointer to normalized current name */
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
char _comp[1024]; /*name component */
size_t nchars; /*characters in component */
size_t offset; /*offset to sym-link value */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_link, FAIL);
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
/* Check args */
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
assert (cur_loc);
assert (new_loc);
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
assert (cur_name && *cur_name);
assert (new_name && *new_name);
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
/* Get normalized copies of the current and new names */
if((norm_cur_name=H5G_normalize(cur_name))==NULL)
HGOTO_ERROR (H5E_SYM, H5E_BADVALUE, FAIL, "can't normalize name");
if((norm_new_name=H5G_normalize(new_name))==NULL)
HGOTO_ERROR (H5E_SYM, H5E_BADVALUE, FAIL, "can't normalize name");
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
switch (type) {
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
case H5G_LINK_SOFT:
/*
* Lookup the the new_name so we can get the group which will contain
* the new entry. The entry shouldn't exist yet.
*/
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
if (H5G_namei(new_loc, norm_new_name, &rest, &grp_ent, NULL,
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
H5G_TARGET_NORMAL, NULL, H5G_NAMEI_TRAVERSE, NULL, dxpl_id)>=0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_EXISTS, FAIL, "already exists");
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
H5E_clear (NULL); /*it's okay that we didn't find it*/
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
rest = H5G_component (rest, &nchars);
/*
* There should be one component left. Make sure it's null
* terminated and that `rest' points to it.
*/
if (rest[nchars]) {
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
if (nchars+1 > sizeof _comp) {
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_COMPLEN, FAIL, "name component is too long");
} else {
HDmemcpy (_comp, rest, nchars);
_comp[nchars] = '\0';
rest = _comp;
}
}
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
/*
* Add the link-value to the local heap for the symbol table which
* will contain the link.
*/
[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
if (NULL==H5O_read (&grp_ent, H5O_STAB_ID, 0, &stab_mesg, dxpl_id))
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_CANTINIT, FAIL, "unable to determine local heap address");
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if ((size_t)(-1)==(offset=H5HL_insert (grp_ent.file, dxpl_id,
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
stab_mesg.heap_addr, HDstrlen(norm_cur_name)+1, norm_cur_name)))
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_CANTINIT, FAIL, "unable to write link value to local heap");
[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
H5O_reset (H5O_STAB_ID, &stab_mesg);
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
/*
* Create a symbol table entry for the link. The object header is
* undefined and the cache contains the link-value offset.
*/
HDmemset (&cur_obj, 0, sizeof cur_obj);
cur_obj.header = HADDR_UNDEF;
cur_obj.file = grp_ent.file;
cur_obj.type = H5G_CACHED_SLINK;
cur_obj.cache.slink.lval_offset = offset;
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
cur_obj_init=1; /* Indicate that the cur_obj struct is initialized */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
/*
* Insert the link head in the symbol table. This shouldn't ever
* fail because we've already checked that the link head doesn't
* exist and the file is writable (because the local heap is
* writable). But if it does, the only side effect is that the local
* heap has some extra garbage in it.
*/
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_stab_insert (&grp_ent, rest, &cur_obj, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_CANTINIT, FAIL, "unable to create new name/link for object");
break;
case H5G_LINK_HARD:
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
if (H5G_namei(cur_loc, norm_cur_name, NULL, NULL, &cur_obj, namei_flags, NULL, H5G_NAMEI_TRAVERSE, NULL, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, FAIL, "source object not found");
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
cur_obj_init=1; /* Indicate that the cur_obj struct is initialized */
if (H5G_insert (new_loc, norm_new_name, &cur_obj, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_CANTINIT, FAIL, "unable to create new name/link for object");
break;
default:
HGOTO_ERROR (H5E_SYM, H5E_BADVALUE, FAIL, "unrecognized link type");
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
}
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Free the group's ID to name buffer, if creating a soft link */
if(type == H5G_LINK_SOFT)
H5G_free_ent_name(&grp_ent);
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Free the ID to name buffer */
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
if(cur_obj_init)
H5G_free_ent_name(&cur_obj);
/* Free the normalized path names */
if(norm_cur_name)
H5MM_xfree(norm_cur_name);
if(norm_new_name)
H5MM_xfree(norm_new_name);
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r337] Changes since 19980403 ---------------------- ./configure.in Moved setting of compiler warning switches earlier in the file. Turned on more warning switches to gcc. ./config/linux Prints a warning if the gcc version is less than 2.8.1 since that version has problems with register allocation for `long long'. ./html/Datatypes.html Documented sharing of data types between datasets. ./src/H5G.c ./src/H5Gpublic.h Implemented H5Gmove(), H5Glink() and H5Gunlink() for hard links. Still have soft links to do. ./src/H5AC.c ./src/H5ACprivate.h ./src/H5D.c ./src/H5E.c ./src/H5Eprivate.h ./src/H5F.c ./src/H5Farray.c ./src/H5Fcore.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fprivate.h ./src/H5Fpublic.h ./src/H5Fsec2.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5HG.c ./src/H5HL.c ./src/H5HLprivate.h ./src/H5I.c ./src/H5Iprivate.h ./src/H5MM.c ./src/H5MMprivate.h ./src/H5O.c ./src/H5Oefl.c ./src/H5Oprivate.h ./src/H5Osdspace.c ./src/H5Oshared.c ./src/H5Ostab.c ./src/H5P.c ./src/H5S.c ./src/H5Ssimp.c ./src/H5T.c ./src/H5Tconv.c ./src/H5Tprivate.h ./src/H5Tpublic.h ./src/H5V.c ./src/H5Vprivate.h ./src/H5detect.c ./src/h5ls.c ./test/cmpd_dset.c ./test/dsets.c ./test/external.c ./test/hyperslab.c ./test/iopipe.c ./test/istore.c ./test/shtype.c ./test/tstab.c Fixed comparisons between signed and unsigned values. Fixed warnings about unused function arguments.
1998-04-07 23:34:16 +08:00
}
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_get_type
*
* Purpose: Returns the type of object pointed to by `ent'.
*
* Return: Success: An object type defined in H5Gpublic.h
*
* Failure: H5G_UNKNOWN
*
* Programmer: Robb Matzke
* Wednesday, November 4, 1998
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
[svn-r7192] Purpose: Code cleanup Description: Clean up various minor warnings in the library. Platforms tested: FreeBSD 4.8 (sleipnir) too small to need h5committest
2003-07-10 03:16:17 +08:00
H5G_obj_t
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
H5G_get_type(H5G_entry_t *ent, hid_t dxpl_id)
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
{
htri_t isa;
size_t i;
[svn-r7192] Purpose: Code cleanup Description: Clean up various minor warnings in the library. Platforms tested: FreeBSD 4.8 (sleipnir) too small to need h5committest
2003-07-10 03:16:17 +08:00
H5G_obj_t ret_value=H5G_UNKNOWN; /* Return value */
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_get_type, H5G_UNKNOWN);
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
for (i=H5G_ntypes_g; i>0; --i) {
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if ((isa=(H5G_type_g[i-1].isa)(ent, dxpl_id))<0) {
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, H5G_UNKNOWN, "unable to determine object type");
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
} else if (isa) {
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_DONE(H5G_type_g[i-1].type);
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
}
}
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (0==i)
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, H5G_UNKNOWN, "unable to determine object type");
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
}
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/*-------------------------------------------------------------------------
[svn-r654] Fixed last bug (hah! ;-) for beta release, changed H5Gget_stat to H5Gget_objinfo and renamed internal functions to match up with the renamed API functions.
1998-09-03 05:16:24 +08:00
* Function: H5G_get_objinfo
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
*
* Purpose: Returns information about an object.
*
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
* Return: Success: Non-negative with info about the object
* returned through STATBUF if it isn't the null
* pointer.
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Failure: Negative
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
*
* Programmer: Robb Matzke
* Monday, April 13, 1998
*
* Modifications:
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Pedro Vicente, <pvn@ncsa.uiuc.edu> 18 Sep 2002
* Added `id to name' support.
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
*
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
*-------------------------------------------------------------------------
*/
herr_t
[svn-r654] Fixed last bug (hah! ;-) for beta release, changed H5Gget_stat to H5Gget_objinfo and renamed internal functions to match up with the renamed API functions.
1998-09-03 05:16:24 +08:00
H5G_get_objinfo (H5G_entry_t *loc, const char *name, hbool_t follow_link,
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
H5G_stat_t *statbuf/*out*/, hid_t dxpl_id)
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
{
H5O_stab_t stab_mesg;
H5G_entry_t grp_ent, obj_ent;
const char *s = NULL;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_get_objinfo, FAIL);
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
assert (loc);
assert (name && *name);
if (statbuf) HDmemset (statbuf, 0, sizeof *statbuf);
/* Find the object's symbol table entry */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if (H5G_namei(loc, name, NULL, &grp_ent/*out*/, &obj_ent/*out*/,
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
(unsigned)(follow_link?H5G_TARGET_NORMAL:H5G_TARGET_SLINK), NULL, H5G_NAMEI_TRAVERSE, NULL, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_NOTFOUND, FAIL, "unable to stat object");
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/*
* Initialize the stat buf. Symbolic links aren't normal objects and
[svn-r437] Changes since 19980612 ---------------------- ./src/H5Tbit.c ./src/H5Tpkg.h Fixed a bug in H5T_bit_copy(). Added H5T_bit_get_d() and H5T_bit_set_d() which treat portions of a bit vector as an unsigned integer. Added H5T_bit_inc() that increments part of a bit vector and returns an indication of overflow. ./src/H5Tconv.c ./src/H5Tpkg.h ./test/dtypes.c Added a slow general floating point conversion which works so far on Intel, MIPS, and DEC but the test is turned off because a cast from double to float will cause a SIGFPE on some systems if an overflow occurs. Added fast hardware conversions between native floating point types. This function is also subject to the SIGFPE problem. ./src/H5detect.c Removed the exponent bias adjustment when the significand isn't normalized. This is now handled in the conversion functions instead. ./src/H5T.c Register new conversion functions. Plugged a memory leak in the library termination code. ./RELEASE Added a list of major changes since the first alpha. ./src/H5.c ./src/H5private.h ./src/H5A.c ./src/H5AC.c ./src/H5D.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5G.c ./src/H5Gprivate.h ./src/H5HG.c ./src/H5O.c ./src/H5T.c ./src/H5Tbit.c ./src/H5Tconv.c Fixed various compiler warnings on Irix64. ./src/H5MM.c Added PABLO_MASK to this file. ./test/chunk.c Removed a warning about memcpy().
1998-07-01 05:30:28 +08:00
* therefore don't have much of the normal info. However, the link value
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
* length is specific to symbolic links.
*/
if (statbuf) {
[svn-r594] Changes since 19980813 ---------------------- ./src/H5G.c ./src/H5Gpublic.h ./test/links.c Fixed a bug in H5Gstat() that caused the wrong information to be returned sometimes. Added check for named data types. ./test/extend.c ./test/links.c ./test/mtime.c Added `all tests passed' messages.
1998-08-15 05:05:16 +08:00
if (H5G_CACHED_SLINK==obj_ent.type) {
[svn-r7498] Purpose: Code cleanup Description: Clean up various warnings and parameter mis-matches, etc. Platforms tested: FreeBSD 4.9 (sleipnir) too minor to need h5committest
2003-09-20 04:13:47 +08:00
const H5HL_t *heap;
[svn-r7489] Purpose: Update Description: A lot of modifications for the FPHDF5 stuff: H5AC.c H5ACprivate.h - Removed AC_find (it's replaced with AC_protect and AC_unprotect). Added flushing if it's an FPHDF5 driver and we're doing an AC_set or AC_unprotect with the dirty flag set. H5B.c - Split up the B_flush function into different functions since the one function was doing serialization which is better left as a separate entity. H5D.c - Removed some FPHDF5 code that was incorrect H5F.c - Split up the F_flush function so that it no longer allocates file space. Created new functions (F_init_superblock, F_read_superblock, and F_write_superblock) for greater modularity and so that the FPHDF5 non-captain processes can read the superblock after the captain process writes it. H5FD.c - Error message correction. H5FDfphdf5.c - Removed MPI barrier call that wasn't needed. H5FPclient.c H5FPserver.c - Modified so that if a process requests data that isn't exactly aligned, we can return it if we have the block that contains the requested address. H5G.c H5Gent.c H5Gnode.c H5HL.c H5HLpkg.h H5HLprivate.h H5Oefl.c - Removed the H5HL_peek function since it was doing a (now unsafe) holding of the information in the cache. Replaced with protect and unprotect calls. H5TB.c - Error fix. The TB_dless function wasn't working properly. H5Gstab.c - Format change. H5err.txt H5Edefin.h H5Einit.h H5Epubgen.h H5Eterm.h - Added new error code. Platforms tested: Modi4 (paralle, Fortran) Sol (Fortran) Linux (C++, Fortran) Copper (Parallel, Fortran) Misc. update:
2003-09-19 03:27:27 +08:00
[svn-r594] Changes since 19980813 ---------------------- ./src/H5G.c ./src/H5Gpublic.h ./test/links.c Fixed a bug in H5Gstat() that caused the wrong information to be returned sometimes. Added check for named data types. ./test/extend.c ./test/links.c ./test/mtime.c Added `all tests passed' messages.
1998-08-15 05:05:16 +08:00
/* Named object is a symbolic link */
[svn-r7489] Purpose: Update Description: A lot of modifications for the FPHDF5 stuff: H5AC.c H5ACprivate.h - Removed AC_find (it's replaced with AC_protect and AC_unprotect). Added flushing if it's an FPHDF5 driver and we're doing an AC_set or AC_unprotect with the dirty flag set. H5B.c - Split up the B_flush function into different functions since the one function was doing serialization which is better left as a separate entity. H5D.c - Removed some FPHDF5 code that was incorrect H5F.c - Split up the F_flush function so that it no longer allocates file space. Created new functions (F_init_superblock, F_read_superblock, and F_write_superblock) for greater modularity and so that the FPHDF5 non-captain processes can read the superblock after the captain process writes it. H5FD.c - Error message correction. H5FDfphdf5.c - Removed MPI barrier call that wasn't needed. H5FPclient.c H5FPserver.c - Modified so that if a process requests data that isn't exactly aligned, we can return it if we have the block that contains the requested address. H5G.c H5Gent.c H5Gnode.c H5HL.c H5HLpkg.h H5HLprivate.h H5Oefl.c - Removed the H5HL_peek function since it was doing a (now unsafe) holding of the information in the cache. Replaced with protect and unprotect calls. H5TB.c - Error fix. The TB_dless function wasn't working properly. H5Gstab.c - Format change. H5err.txt H5Edefin.h H5Einit.h H5Epubgen.h H5Eterm.h - Added new error code. Platforms tested: Modi4 (paralle, Fortran) Sol (Fortran) Linux (C++, Fortran) Copper (Parallel, Fortran) Misc. update:
2003-09-19 03:27:27 +08:00
if (NULL == H5O_read(&grp_ent, H5O_STAB_ID, 0, &stab_mesg, dxpl_id))
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, FAIL, "unable to read symbolic link value")
if (NULL == (heap = H5HL_protect(grp_ent.file, dxpl_id, stab_mesg.heap_addr)))
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, FAIL, "unable to read protect link value")
s = H5HL_offset_into(grp_ent.file, heap, obj_ent.cache.slink.lval_offset);
statbuf->linklen = HDstrlen(s) + 1; /*count the null terminator*/
if (H5HL_unprotect(grp_ent.file, dxpl_id, heap, stab_mesg.heap_addr) < 0)
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, FAIL, "unable to read unprotect link value")
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
statbuf->objno = 0;
[svn-r594] Changes since 19980813 ---------------------- ./src/H5G.c ./src/H5Gpublic.h ./test/links.c Fixed a bug in H5Gstat() that caused the wrong information to be returned sometimes. Added check for named data types. ./test/extend.c ./test/links.c ./test/mtime.c Added `all tests passed' messages.
1998-08-15 05:05:16 +08:00
statbuf->nlink = 0;
statbuf->type = H5G_LINK;
statbuf->mtime = 0;
} else {
/* Some other type of object */
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
statbuf->objno = obj_ent.header;
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
statbuf->nlink = H5O_link (&obj_ent, 0, dxpl_id);
[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
if (NULL==H5O_read(&obj_ent, H5O_MTIME_ID, 0, &(statbuf->mtime), dxpl_id)) {
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
H5E_clear(NULL);
[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
if (NULL==H5O_read(&obj_ent, H5O_MTIME_NEW_ID, 0, &(statbuf->mtime), dxpl_id)) {
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
H5E_clear(NULL);
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
statbuf->mtime = 0;
}
[svn-r5110] ./hdf5-devel/src/H5G.c Merge of name component length limit bugfix from 1.4 branch. A couple of very minor typos fixed.
2002-03-29 03:27:38 +08:00
}
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
statbuf->type = H5G_get_type(&obj_ent, dxpl_id);
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
H5E_clear(NULL); /*clear errors resulting from checking type*/
[svn-r7561] Purpose: Feature add Description: Add a few new fields to the H5G_stat_t structure, to allow more information about the object header to be retrieved. Platforms tested: FreeBSD 4.9 (sleipnir) h5committest
2003-10-07 23:27:19 +08:00
/* Get object header information */
if(H5O_get_info(&obj_ent, &(statbuf->ohdr), dxpl_id)<0)
HGOTO_ERROR(H5E_SYM, H5E_CANTGET, FAIL, "unable to get object header information")
[svn-r5110] ./hdf5-devel/src/H5G.c Merge of name component length limit bugfix from 1.4 branch. A couple of very minor typos fixed.
2002-03-29 03:27:38 +08:00
}
[svn-r5096] Purpose: Bug Fix Description: The H5Gget_objinfo() function was not setting the 'fileno' field in the H5G_stat_t struct passed in. Solution: Added a "file serial number" to each file currently open in the library and put that in the 'fileno' field. If a file is opened twice (with H5Fopen) and the VFL driver detects that it is the same file (i.e. the two file structures have the same "shared file info" in the library's memory structures), they will have the same serial number. This serial number has two drawbacks: - If a VFL driver doesn't/can't detect that two calls to H5Fopen with the same file actually _are_ the same file, each will get a different serial number - If the same file is closed and re-opened, the serial number will be different. It is be possible to fix the second drawback for many VFL drivers, but it would be a lot of effort and probably isn't worth it until we've got a good reason to do it. Dunno if we'll ever be able to fix the first drawback... Platforms tested: FreeBSD 4.5 (sleipnir) VS: ----------------------------------------------------------------------
2002-03-28 04:25:03 +08:00
/* Common code to retrieve the file's fileno */
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
if(H5F_get_fileno(obj_ent.file,&statbuf->fileno)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_FILE, H5E_BADVALUE, FAIL, "unable to read fileno");
[svn-r5110] ./hdf5-devel/src/H5G.c Merge of name component length limit bugfix from 1.4 branch. A couple of very minor typos fixed.
2002-03-29 03:27:38 +08:00
}
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Free the ID to name buffers */
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
H5G_free_ent_name(&grp_ent);
H5G_free_ent_name(&obj_ent);
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
}
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_get_num_objs
*
* Purpose: Private function for H5Gget_num_objs. Returns the number
* of objects in the group. It iterates all B-tree leaves
* and sum up total number of group members.
*
* Return: Success: Non-negative
*
* Failure: Negative
*
* Programmer: Raymond Lu
* Nov 20, 2002
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
static herr_t
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
H5G_get_num_objs(H5G_entry_t *loc, hsize_t *num_objs, hid_t dxpl_id)
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
{
herr_t ret_value;
FUNC_ENTER_NOAPI(H5G_get_num_objs, FAIL);
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
/* Sanity check */
assert(loc);
assert(loc->type==H5G_CACHED_STAB);
assert(num_objs);
/* Reset the number of objects in the group */
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
*num_objs = 0;
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
/* Iterate over the group members */
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
if ((ret_value = H5B_iterate (loc->file, dxpl_id, H5B_SNODE,
H5G_node_sumup, loc->cache.stab.btree_addr, num_objs))<0)
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
HERROR (H5E_SYM, H5E_CANTINIT, "iteration operator failed");
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
}
/*-------------------------------------------------------------------------
* Function: H5G_get_objname_by_idx
*
* Purpose: Private function for H5Gget_objname_by_idx.
* Returns the name of objects in the group by giving index.
*
* Return: Success: Non-negative
*
* Failure: Negative
*
* Programmer: Raymond Lu
* Nov 20, 2002
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
static ssize_t
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
H5G_get_objname_by_idx(H5G_entry_t *loc, hsize_t idx, char* name, size_t size, hid_t dxpl_id)
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
{
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
H5G_bt_ud3_t udata; /* Iteration information */
ssize_t ret_value; /* Return value */
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
FUNC_ENTER_NOAPI(H5G_get_objname_by_idx, FAIL);
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
/* Sanity check */
assert(loc);
assert(loc->type==H5G_CACHED_STAB);
/* Set iteration information */
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
udata.idx = idx;
udata.num_objs = 0;
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
udata.ent = loc;
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
udata.name = NULL;
/* Iterate over the group members */
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
if ((ret_value = H5B_iterate (loc->file, dxpl_id, H5B_SNODE,
H5G_node_name, loc->cache.stab.btree_addr, &udata))<0)
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, FAIL, "iteration operator failed");
[svn-r7412] Purpose: Bug fix Description: H5Gget_objname_by_idx should allow NULL for the 'name' parameter, to allow for querying for the name's length. Solution: Allow NULL for the 'name' parameter, the internal functions were already ready to handle that case. Clean up RM information about H5Gget_objname_by_idx and H5Gget_objtype_by_idx also. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-27 02:35:37 +08:00
/* Get the length of the name */
[svn-r6843] Purpose: Code cleanup Description: Clean up warnings exposed by compiling on O2K. Also, revert some of Bill and my changes to the H5S_mpi_opt_types_g, etc. and settle them back into their original location. Platforms tested: h5committested.
2003-05-10 02:18:21 +08:00
ret_value = (ssize_t)HDstrlen(udata.name);
[svn-r7412] Purpose: Bug fix Description: H5Gget_objname_by_idx should allow NULL for the 'name' parameter, to allow for querying for the name's length. Solution: Allow NULL for the 'name' parameter, the internal functions were already ready to handle that case. Clean up RM information about H5Gget_objname_by_idx and H5Gget_objtype_by_idx also. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-27 02:35:37 +08:00
/* Copy the name into the user's buffer, if given */
if(name) {
HDstrncpy(name, udata.name, MIN((size_t)(ret_value+1),size));
[svn-r6142] Purpose: Code cleanup Description: Clean up warnings with gcc 3.2.1 Platforms tested: FreeBSD 4.7 (sleipnir) w/C++, changes too small for triple check
2002-12-02 21:12:17 +08:00
if((size_t)ret_value >= size)
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
name[size-1]='\0';
[svn-r7412] Purpose: Bug fix Description: H5Gget_objname_by_idx should allow NULL for the 'name' parameter, to allow for querying for the name's length. Solution: Allow NULL for the 'name' parameter, the internal functions were already ready to handle that case. Clean up RM information about H5Gget_objname_by_idx and H5Gget_objtype_by_idx also. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-27 02:35:37 +08:00
} /* end if */
[svn-r6142] Purpose: Code cleanup Description: Clean up warnings with gcc 3.2.1 Platforms tested: FreeBSD 4.7 (sleipnir) w/C++, changes too small for triple check
2002-12-02 21:12:17 +08:00
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
done:
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/* Free the duplicated name */
if(udata.name!=NULL)
H5MM_xfree(udata.name);
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
}
/*-------------------------------------------------------------------------
* Function: H5G_get_objtype_by_idx
*
* Purpose: Private function for H5Gget_objtype_by_idx.
* Returns the type of objects in the group by giving index.
*
* Return: Success: H5G_GROUP(1), H5G_DATASET(2), H5G_TYPE(3)
*
* Failure: UNKNOWN
*
* Programmer: Raymond Lu
* Nov 20, 2002
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
[svn-r7192] Purpose: Code cleanup Description: Clean up various minor warnings in the library. Platforms tested: FreeBSD 4.8 (sleipnir) too small to need h5committest
2003-07-10 03:16:17 +08:00
static H5G_obj_t
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
H5G_get_objtype_by_idx(H5G_entry_t *loc, hsize_t idx, hid_t dxpl_id)
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
{
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
H5G_bt_ud3_t udata; /* User data for B-tree callback */
H5G_obj_t ret_value; /* Return value */
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
[svn-r7638] Purpose: Code cleanup Description: Clean up compiler warnings Platforms tested: h5committest
2003-10-16 04:02:51 +08:00
FUNC_ENTER_NOAPI(H5G_get_objtype_by_idx, H5G_UNKNOWN);
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
/* Sanity check */
assert(loc);
assert(loc->type==H5G_CACHED_STAB);
/* Set iteration information */
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
udata.idx = idx;
udata.num_objs = 0;
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
udata.ent = loc;
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
/* Iterate over the group members */
[svn-r7391] Purpose: Bug fix Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx were only accepting a group ID, instead of a location ID, as our documentation for them stated. Solution: Allow them to accept a location ID. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-08-22 21:50:01 +08:00
if (H5B_iterate (loc->file, dxpl_id, H5B_SNODE,
H5G_node_type, loc->cache.stab.btree_addr, &udata)<0)
[svn-r7638] Purpose: Code cleanup Description: Clean up compiler warnings Platforms tested: h5committest
2003-10-16 04:02:51 +08:00
HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, H5G_UNKNOWN, "iteration operator failed");
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
ret_value = udata.type;
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r6136] Purpose: new functions Description: H5Gget_num_objs, H5Gget_objname_by_idx and H5Gget_objtype_by_idx. Platforms tested: modi4, arabica, eirene Misc. update: RELEASE.txt updated.
2002-11-26 01:59:14 +08:00
}
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_linkval
*
* Purpose: Returns the value of a symbolic link.
*
[svn-r876] Changes since 19981102 ---------------------- ./bin/snapshot Made same fix as for the release script yesterday. ./src/H5D.c ./src/H5Dprivate.h ./src/H5G.c ./src/H5Gprivate.h ./src/H5Gpublic.h ./src/H5O.c ./src/H5Oprivate.h ./src/H5RA.c ./src/H5RAprivate.h ./src/H5T.c ./src/H5Tprivate.h Improved object type checking. Instead of determining the object type by trying to open each of the possible types, we keep a table of associations between object type number (like H5G_GROUP, H5G_DATASET, H5D_TYPE, and H5D_RAGGED) and an `isa' function that returns true if the object header has the right messages to make the object a particular type. This mechanism also allows specialization of object types by permitting an object to satisfy more than one `isa' function. Added `isa' functions for groups, datasets, ragged arrays, and committed data types. ./src/H5config.h.in Added HAVE_STAT_ST_BLOCKS. I thought this had already been added, but apparently not. ./tools/h5ls.c Removed system include files since they're already included by H5private.h and since I wasn't including them portably anyway. By default, 1-byte integer types are printed as integer values instead of ASCII characters. However, the `-s' or `--string' command-line switch causes the data to be interpretted as ASCII. String data types are always printed as character data. Ragged arrays are now identified as ragged arrays and h5ls doesn't descend into the group automatically. This uses the new object type specialization stuff. ./tools/h5tools.c ./tools/h5tools.h Added the ability to print 1-byte integer types as either ASCII or numeric data instead of always ASCII. The default is to print as numeric data.
1998-11-06 04:28:34 +08:00
* Return: Success: Non-negative, with at most SIZE bytes of the
* link value copied into the BUF buffer. If the
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
* link value is larger than SIZE characters
* counting the null terminator then the BUF
* result will not be null terminated.
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Failure: Negative
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
*
* Programmer: Robb Matzke
* Monday, April 13, 1998
*
* Modifications:
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Pedro Vicente, <pvn@ncsa.uiuc.edu> 18 Sep 2002
* Added `id to name' support.
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
*
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
*-------------------------------------------------------------------------
*/
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
static herr_t
H5G_linkval (H5G_entry_t *loc, const char *name, size_t size, char *buf/*out*/, hid_t dxpl_id)
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
{
const char *s = NULL;
H5G_entry_t grp_ent, obj_ent;
H5O_stab_t stab_mesg;
[svn-r7498] Purpose: Code cleanup Description: Clean up various warnings and parameter mis-matches, etc. Platforms tested: FreeBSD 4.9 (sleipnir) too minor to need h5committest
2003-09-20 04:13:47 +08:00
const H5HL_t *heap;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_linkval, FAIL);
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/*
* Get the symbol table entry for the link head and the symbol table
* entry for the group in which the link head appears.
*/
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if (H5G_namei(loc, name, NULL, &grp_ent/*out*/, &obj_ent/*out*/,
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
H5G_TARGET_SLINK, NULL, H5G_NAMEI_TRAVERSE, NULL, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_NOTFOUND, FAIL, "symbolic link was not found");
if (H5G_CACHED_SLINK!=obj_ent.type)
HGOTO_ERROR (H5E_SYM, H5E_NOTFOUND, FAIL, "object is not a symbolic link");
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/*
* Get the address of the local heap for the link value and a pointer
* into that local heap.
*/
[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
if (NULL==H5O_read (&grp_ent, H5O_STAB_ID, 0, &stab_mesg, dxpl_id))
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR (H5E_SYM, H5E_CANTINIT, FAIL, "unable to determine local heap address");
[svn-r7489] Purpose: Update Description: A lot of modifications for the FPHDF5 stuff: H5AC.c H5ACprivate.h - Removed AC_find (it's replaced with AC_protect and AC_unprotect). Added flushing if it's an FPHDF5 driver and we're doing an AC_set or AC_unprotect with the dirty flag set. H5B.c - Split up the B_flush function into different functions since the one function was doing serialization which is better left as a separate entity. H5D.c - Removed some FPHDF5 code that was incorrect H5F.c - Split up the F_flush function so that it no longer allocates file space. Created new functions (F_init_superblock, F_read_superblock, and F_write_superblock) for greater modularity and so that the FPHDF5 non-captain processes can read the superblock after the captain process writes it. H5FD.c - Error message correction. H5FDfphdf5.c - Removed MPI barrier call that wasn't needed. H5FPclient.c H5FPserver.c - Modified so that if a process requests data that isn't exactly aligned, we can return it if we have the block that contains the requested address. H5G.c H5Gent.c H5Gnode.c H5HL.c H5HLpkg.h H5HLprivate.h H5Oefl.c - Removed the H5HL_peek function since it was doing a (now unsafe) holding of the information in the cache. Replaced with protect and unprotect calls. H5TB.c - Error fix. The TB_dless function wasn't working properly. H5Gstab.c - Format change. H5err.txt H5Edefin.h H5Einit.h H5Epubgen.h H5Eterm.h - Added new error code. Platforms tested: Modi4 (paralle, Fortran) Sol (Fortran) Linux (C++, Fortran) Copper (Parallel, Fortran) Misc. update:
2003-09-19 03:27:27 +08:00
if (NULL == (heap = H5HL_protect(grp_ent.file, dxpl_id, stab_mesg.heap_addr)))
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, FAIL, "unable to read protect link value")
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r7489] Purpose: Update Description: A lot of modifications for the FPHDF5 stuff: H5AC.c H5ACprivate.h - Removed AC_find (it's replaced with AC_protect and AC_unprotect). Added flushing if it's an FPHDF5 driver and we're doing an AC_set or AC_unprotect with the dirty flag set. H5B.c - Split up the B_flush function into different functions since the one function was doing serialization which is better left as a separate entity. H5D.c - Removed some FPHDF5 code that was incorrect H5F.c - Split up the F_flush function so that it no longer allocates file space. Created new functions (F_init_superblock, F_read_superblock, and F_write_superblock) for greater modularity and so that the FPHDF5 non-captain processes can read the superblock after the captain process writes it. H5FD.c - Error message correction. H5FDfphdf5.c - Removed MPI barrier call that wasn't needed. H5FPclient.c H5FPserver.c - Modified so that if a process requests data that isn't exactly aligned, we can return it if we have the block that contains the requested address. H5G.c H5Gent.c H5Gnode.c H5HL.c H5HLpkg.h H5HLprivate.h H5Oefl.c - Removed the H5HL_peek function since it was doing a (now unsafe) holding of the information in the cache. Replaced with protect and unprotect calls. H5TB.c - Error fix. The TB_dless function wasn't working properly. H5Gstab.c - Format change. H5err.txt H5Edefin.h H5Einit.h H5Epubgen.h H5Eterm.h - Added new error code. Platforms tested: Modi4 (paralle, Fortran) Sol (Fortran) Linux (C++, Fortran) Copper (Parallel, Fortran) Misc. update:
2003-09-19 03:27:27 +08:00
s = H5HL_offset_into(grp_ent.file, heap, obj_ent.cache.slink.lval_offset);
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
/* Copy to output buffer */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (size>0 && buf)
[svn-r673] Changes since 19980908 ---------------------- ./COPYING Reformatted as text instead of C. Removed zlib crew from the list of contributors since no zlib code is actually in the hdf5 library. ./INSTALL ./INSTALL.ascired ./MANIFEST Minor updates for Beta release including version number change. ./INSTALL_MAINT Added information about making a release. ./RELEASE Updated function list based on public header files. ./bin/checkposix Got rid of complaints about some obvious things. ./doc/html/H5.api.html ./doc/html/RM_H5F.html ./src/H5F.c ./src/H5Fpublic.h ./test/tfile.c Changed H5Fget_create_template() and H5Fget_access_template() to H5Fget_create_plist() and H5Fget_access_plist() since that conforms better to lots of other names. ./doc/html/Datatypes.html ./doc/html/ExternalFiles.html ./doc/html/Files.html ./doc/html/H5.api.html ./doc/html/H5.sample_code.html ./doc/html/RM_H5F.html ./doc/html/RM_H5Front.html Changed `template' to `property list', etc. ./doc/html/Ragged.html [NEW] Documentation for ragged arrays. ./src/H5Iprivate.h ./src/H5Ipublic.h ./src/H5I.c Changed the scope of some symbols to be more local. ./src/H5.c ./src/H5AC.c ./src/H5D.c ./src/H5E.c ./src/H5F.c ./src/H5Ffamily.c ./src/H5Fistore.c ./src/H5Flow.c ./src/H5Fsec2.c ./src/H5Fsplit.c ./src/H5Fstdio.c ./src/H5G.c ./src/H5Gnode.c ./src/H5HG.c ./src/H5I.c ./src/H5O.c ./src/H5Ocomp.c ./src/H5Odtype.c ./src/H5Oefl.c ./src/H5Omtime.c ./src/H5Oname.c ./src/H5P.c ./src/H5S.c ./src/H5Shyper.c ./src/H5Tbit.c ./src/H5Tconv.c ./src/H5V.c ./src/H5Z.c ./src/H5private.h Fixed some violations of our naming scheme by adding HD to the beginning of all Posix functions.
1998-09-09 03:15:44 +08:00
HDstrncpy (buf, s, size);
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r7489] Purpose: Update Description: A lot of modifications for the FPHDF5 stuff: H5AC.c H5ACprivate.h - Removed AC_find (it's replaced with AC_protect and AC_unprotect). Added flushing if it's an FPHDF5 driver and we're doing an AC_set or AC_unprotect with the dirty flag set. H5B.c - Split up the B_flush function into different functions since the one function was doing serialization which is better left as a separate entity. H5D.c - Removed some FPHDF5 code that was incorrect H5F.c - Split up the F_flush function so that it no longer allocates file space. Created new functions (F_init_superblock, F_read_superblock, and F_write_superblock) for greater modularity and so that the FPHDF5 non-captain processes can read the superblock after the captain process writes it. H5FD.c - Error message correction. H5FDfphdf5.c - Removed MPI barrier call that wasn't needed. H5FPclient.c H5FPserver.c - Modified so that if a process requests data that isn't exactly aligned, we can return it if we have the block that contains the requested address. H5G.c H5Gent.c H5Gnode.c H5HL.c H5HLpkg.h H5HLprivate.h H5Oefl.c - Removed the H5HL_peek function since it was doing a (now unsafe) holding of the information in the cache. Replaced with protect and unprotect calls. H5TB.c - Error fix. The TB_dless function wasn't working properly. H5Gstab.c - Format change. H5err.txt H5Edefin.h H5Einit.h H5Epubgen.h H5Eterm.h - Added new error code. Platforms tested: Modi4 (paralle, Fortran) Sol (Fortran) Linux (C++, Fortran) Copper (Parallel, Fortran) Misc. update:
2003-09-19 03:27:27 +08:00
if (H5HL_unprotect(grp_ent.file, dxpl_id, heap, stab_mesg.heap_addr) < 0)
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, FAIL, "unable to read unprotect link value")
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Free the ID to name buffers */
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
H5G_free_ent_name(&grp_ent);
H5G_free_ent_name(&obj_ent);
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
}
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_set_comment
*
* Purpose: (Re)sets the comment for an object.
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Return: Non-negative on success/Negative on failure
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
*
* Programmer: Robb Matzke
* Monday, July 20, 1998
*
* Modifications:
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Pedro Vicente, <pvn@ncsa.uiuc.edu> 18 Sep 2002
* Added `id to name' support.
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
*
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
*-------------------------------------------------------------------------
*/
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
static herr_t
H5G_set_comment(H5G_entry_t *loc, const char *name, const char *buf, hid_t dxpl_id)
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
{
H5G_entry_t obj_ent;
H5O_name_t comment;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_set_comment, FAIL);
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
/* Get the symbol table entry for the object */
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
if (H5G_namei(loc, name, NULL, NULL, &obj_ent/*out*/, H5G_TARGET_NORMAL,
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
NULL, H5G_NAMEI_TRAVERSE, NULL, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, FAIL, "object not found");
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
/* Remove the previous comment message if any */
[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
if (H5O_remove(&obj_ent, H5O_NAME_ID, 0, dxpl_id)<0)
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
H5E_clear(NULL);
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
/* Add the new message */
if (buf && *buf) {
comment.s = H5MM_xstrdup(buf);
[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
if (H5O_modify(&obj_ent, H5O_NAME_ID, H5O_NEW_MESG, 0, 1, &comment, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_OHDR, H5E_CANTINIT, FAIL, "unable to set comment object header message");
[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
H5O_reset(H5O_NAME_ID, &comment);
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
}
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Free the ID to name buffer */
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
H5G_free_ent_name(&obj_ent);
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
}
/*-------------------------------------------------------------------------
* Function: H5G_get_comment
*
* Purpose: Get the comment value for an object.
*
[svn-r537] Changes since 19980722 ---------------------- ./src/H5A.c ./src/H5Apublic.h ./test/tattr.c Switched the order of the second and third argument of H5Aget_name() to make it consistent with other functions that take buffers and buffer sizes. ./src/H5G.c ./src/H5Gpublic.h ./src/H5Gprivate.h The H5Gget_comment() function returns the size of the comment including the null terminator. If the object has no comment then zero is returned. If an error occurs then a negative value is returned. ./MANIFEST ./tools/Makefile.in ./tools/h5tools.h [NEW] ./tools/h5dump.c [NEW] Created a library for printing values of datasets in a way that looks nice. It's not done yet, but I needed it for debugging the contents of files from Jim Reus. ./tools/h5ls.c Added the `-d' and `--dump' options which cause the contents of a dataset to be printed. Added `-w N' and `--width=N' options to control how wide the raw data output should be. If you want single-column output then say `-w1'. Printing dataset values can now handle datasets of any integer or floating point atomic type. As a special case, integers which are one byte wide are treated a character strings for now. Sample output: $ h5ls --dump --width=60 banana.hdf ARCHIVE 0:0:0:744 Dataset {52/Inf} Data: (0) "U struct complex { double R; double I; };\012V" (43) " double;\012" U 0:0:0:2500 Dataset {256/512} Data: printing of compound data types is not implemented yet V 0:0:0:3928 Dataset {256/512} Data: (0) 0, 0.015625, 0.03125, 0.046875, 0.0625, (5) 0.078125, 0.09375, 0.109375, 0.125, 0.140625, (10) 0.15625, 0.171875, 0.1875, 0.203125, 0.21875, (15) 0.234375, 0.25, 0.265625, 0.28125, 0.296875, ...
1998-07-24 05:19:17 +08:00
* Return: Success: Number of bytes in the comment including the
* null terminator. Zero if the object has no
* comment.
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Failure: Negative
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
*
* Programmer: Robb Matzke
* Monday, July 20, 1998
*
* Modifications:
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Pedro Vicente, <pvn@ncsa.uiuc.edu> 18 Sep 2002
* Added `id to name' support.
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
*
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
*-------------------------------------------------------------------------
*/
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
static int
H5G_get_comment(H5G_entry_t *loc, const char *name, size_t bufsize, char *buf, hid_t dxpl_id)
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
{
H5O_name_t comment;
H5G_entry_t obj_ent;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
int ret_value;
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_get_comment, FAIL);
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
/* Get the symbol table entry for the object */
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
if (H5G_namei(loc, name, NULL, NULL, &obj_ent/*out*/, H5G_TARGET_NORMAL,
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
NULL, H5G_NAMEI_TRAVERSE, NULL, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, FAIL, "object not found");
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
/* Get the message */
comment.s = NULL;
[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
if (NULL==H5O_read(&obj_ent, H5O_NAME_ID, 0, &comment, dxpl_id)) {
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (buf && bufsize>0)
buf[0] = '\0';
ret_value = 0;
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
} else {
[svn-r6041] Purpose: Bug fix Description: #340 - get comment length for H5G; #435 - H5Aget_storage_size; #644 - H5Arename Platforms tested: eirene, arabica
2002-10-30 00:37:49 +08:00
if(buf && bufsize)
HDstrncpy(buf, comment.s, bufsize);
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
ret_value = (int)HDstrlen(comment.s);
[svn-r6411] Purpose: Code cleanup Description: Clean up miscellaneous warnings which have crept into the code. Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD. Adjust gcc compiler flags to be more concise for production mode. Refactor the H5O code so that there is a stronger boundary between code in the H5O package and code in the library which just calls H5O routines. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2 Misc. update: Update MANIFEST if you add or remove any file.
2003-02-17 23:54:15 +08:00
H5O_reset(H5O_NAME_ID, &comment);
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
}
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Free the ID to name buffer */
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
H5G_free_ent_name(&obj_ent);
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r522] Changes since 19980720 ---------------------- ./src/H5Gpublic.h ./src/H5Gprivate.h ./src/H5G.c ./test/dsets.c ./doc/html/Groups.html Added the H5Gset_comment() and H5Gget_comment() functions described in an earlier e-mail. ./src/H5.c Fixed a bug in the tracing code that caused certain data space enum types to not be printed. Nested case statements can get confusing to the eyes! ./tools/h5ls.c Prints the first 50 bytes or so of object comments.
1998-07-21 05:01:32 +08:00
}
[svn-r346] Changes since 19980410 ---------------------- ./html/H5.format.html ./src/H5E.c ./src/H5Epublic.h ./src/H5F.c ./src/H5G.c ./src/H5Gent.c ./src/H5Gnode.c ./src/H5Gprivate.h ./src/h5ls.c ./test/Makefile.in Symbolic links are now supported. The ./test/links program will create a `links.h5' file that demonstrates hard links, soft links, dangling links, and recursive links. A symbolic link is a symbol table entity and doesn't have an object header. It's format is described in H5.format.hml. ./src/H5G.c ./src/H5Gpublic.h ./src/h5ls.c Implemented H5Gstat() and H5Gget_linkval() as documented by Quincey. The `H5G_type_t type' field of `H5G_stat_t' was changed to `int type' because H5G_type_t was already used and the `type' data type should be open-ended. ./src/H5Ffamily.c Removed an incorrect diagnostic message. ./test/big.c Added read/write calls to test partial I/O to big contiguous datasets. With no arguments it writes to a dataset and prints the list of points written which should be redirected to a file. With an argument (the name of a file containing the stdout of a run with no arguments) values are read from the dataset. One would typically say `big >x && big x'.
1998-04-15 00:44:46 +08:00
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_unlink
*
* Purpose: Unlink a name from a group.
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Return: Non-negative on success/Negative on failure
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
*
* Programmer: Robb Matzke
* Thursday, September 17, 1998
*
* Modifications:
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Pedro Vicente, <pvn@ncsa.uiuc.edu> 18 Sep 2002
* Added `id to name' support.
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
*
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
*-------------------------------------------------------------------------
*/
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
static herr_t
H5G_unlink(H5G_entry_t *loc, const char *name, hid_t dxpl_id)
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
{
H5G_entry_t grp_ent, obj_ent;
size_t len;
const char *base=NULL;
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
char *norm_name = NULL; /* Pointer to normalized name */
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
H5G_stat_t statbuf; /* Info about object to unlink */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
H5RS_str_t *name_r; /* Ref-counted version of name */
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_unlink, FAIL);
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
assert(loc);
assert(name && *name);
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
/* Get normalized copy of the name */
if((norm_name=H5G_normalize(name))==NULL)
HGOTO_ERROR (H5E_SYM, H5E_BADVALUE, FAIL, "can't normalize name");
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Reset the group entries to known values in a portable way */
HDmemset(&grp_ent,0,sizeof(H5G_entry_t));
HDmemset(&obj_ent,0,sizeof(H5G_entry_t));
/* Get object type before unlink */
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
if (H5G_get_objinfo(loc, norm_name, FALSE, &statbuf, dxpl_id)<0)
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, FAIL, "object not found");
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
/* Get the entry for the group that contains the object to be unlinked */
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
if (H5G_namei(loc, norm_name, NULL, &grp_ent, &obj_ent,
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
H5G_TARGET_SLINK|H5G_TARGET_MOUNT, NULL, H5G_NAMEI_TRAVERSE, NULL, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, FAIL, "object not found");
if (!H5F_addr_defined(grp_ent.header))
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, FAIL, "no containing group specified");
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
if (NULL==(base=H5G_basename(norm_name, &len)) || '/'==*base)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, FAIL, "problems obtaining object base name");
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
/* Remove the name from the symbol table */
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_stab_remove(&grp_ent, base, dxpl_id)<0)
[svn-r6654] Purpose: Code cleanup Description: Stop using hard-coded constant for the initial size of the root group's symbol table local heap and use macro. Also, changed existing (unused) macro for the initial heap size to be the same as the hard-coded constant. Platforms tested: FreeBSD 4.8 (sleipnir) w/C++ Linux 2.4 (burrwhite) w/FORTRAN Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/parallel & FORTRAN (h5committest not run due to my ongoing difficulties with C++ on burrwhite).
2003-04-14 12:49:26 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTDELETE, FAIL, "unable to unlink name from symbol table");
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Search the open IDs and replace names for unlinked object */
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
name_r=H5RS_wrap(norm_name);
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
assert(name_r);
if (H5G_replace_name(statbuf.type, &obj_ent, name_r, NULL, NULL, NULL, OP_UNLINK )<0)
[svn-r6654] Purpose: Code cleanup Description: Stop using hard-coded constant for the initial size of the root group's symbol table local heap and use macro. Also, changed existing (unused) macro for the initial heap size to be the same as the hard-coded constant. Platforms tested: FreeBSD 4.8 (sleipnir) w/C++ Linux 2.4 (burrwhite) w/FORTRAN Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/parallel & FORTRAN (h5committest not run due to my ongoing difficulties with C++ on burrwhite).
2003-04-14 12:49:26 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTDELETE, FAIL, "unable to replace name");
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
H5RS_decr(name_r);
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
done:
/* Free the ID to name buffers */
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
H5G_free_ent_name(&grp_ent);
H5G_free_ent_name(&obj_ent);
[svn-r7377] Purpose: Bug fix Description: Linking and unlinking objects with multiple internal and/or trailing '/'s was not handled well in the library. Solution: "Normalize" strings by removing trailing '/'s and collapsing multiple internal '/'s down into just one '/' before operating on the name. Platforms tested: FreeBSD 4.8 (sleipnir) h5committested
2003-08-17 02:54:40 +08:00
/* Free the normalized path name */
if(norm_name)
H5MM_xfree(norm_name);
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r714] Changes since 19980911 ---------------------- This checkin is to fix a couple bugs for Jim Reus. Some features are not fully implemented but it shouldn't break anything. ./config/conclude.in ./test/Makefile.in `make clean' removes object files from the test directory as well as a couple more *.h5 temp files. ./config/hpux9.03 [NEW] New config file for HP/UX 9.03 ./src/H5B.c ./src/H5Bprivate.h ./src/H5Fistore.c ./src/H5G.c ./src/H5Gnode.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5Gstab.c Not-yet-complete version of object removal. ./src/H5Fistore.c ./src/H5Fprivate.h Experimental optimizations, disabled by default. ./src/H5Fprivate.h Default low-level driver was changed to H5F_LOW_SEC instead of H5F_LOW_STDIO because the sec2 driver is much easier to debug. ./src/H5Fsplit.c ./src/H5G.c ./src/H5Z.c Changed a couple return statements to HRETURN. ./src/H5Odtype.c Removed a check for nested compound data types from back when they weren't implemented that raised an error. ./tools/h5tools.c Increased temp buffer sizes and added a check for buffer overflow so we fail an assertion (hopefully). This really needs to be fixed sometime. Added support for printing H5T_STRING data types.
1998-09-22 23:27:26 +08:00
}
[svn-r726] Changes since 19980924 ---------------------- ./MANIFEST ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./test/Makefile.in ./test/unlink.c [NEW] Finished H5Gunlink() and H5Grename(). ./src/H5F.c ./src/H5Fistore.c ./src/H5Fprivate.h Removed the last memcpy() from the chunk cache. ./src/H5Fistore.c The offset of a chunk within a dataset is an 8-byte quantity per dimension instead of 4 bytes. ./src/H5HL.c Fixed infinite loops in H5HL_remove().
1998-09-28 22:20:21 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_move
*
* Purpose: Atomically rename an object.
*
[svn-r797] Changed comments from returning "SUCCEED/FAIL" to "non-negative/negative" and also fixed a few more explicit checks against FAIL.
1998-10-27 05:18:54 +08:00
* Return: Non-negative on success/Negative on failure
[svn-r726] Changes since 19980924 ---------------------- ./MANIFEST ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./test/Makefile.in ./test/unlink.c [NEW] Finished H5Gunlink() and H5Grename(). ./src/H5F.c ./src/H5Fistore.c ./src/H5Fprivate.h Removed the last memcpy() from the chunk cache. ./src/H5Fistore.c The offset of a chunk within a dataset is an 8-byte quantity per dimension instead of 4 bytes. ./src/H5HL.c Fixed infinite loops in H5HL_remove().
1998-09-28 22:20:21 +08:00
*
* Programmer: Robb Matzke
* Friday, September 25, 1998
*
* Modifications:
*
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
* Raymond Lu
* Thursday, April 18, 2002
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Pedro Vicente, <pvn@ncsa.uiuc.edu> 22 Aug 2002
* Added `id to name' support.
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
*
[svn-r726] Changes since 19980924 ---------------------- ./MANIFEST ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./test/Makefile.in ./test/unlink.c [NEW] Finished H5Gunlink() and H5Grename(). ./src/H5F.c ./src/H5Fistore.c ./src/H5Fprivate.h Removed the last memcpy() from the chunk cache. ./src/H5Fistore.c The offset of a chunk within a dataset is an 8-byte quantity per dimension instead of 4 bytes. ./src/H5HL.c Fixed infinite loops in H5HL_remove().
1998-09-28 22:20:21 +08:00
*-------------------------------------------------------------------------
*/
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
static herr_t
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
H5G_move(H5G_entry_t *src_loc, const char *src_name, H5G_entry_t *dst_loc,
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
const char *dst_name, hid_t dxpl_id)
[svn-r726] Changes since 19980924 ---------------------- ./MANIFEST ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./test/Makefile.in ./test/unlink.c [NEW] Finished H5Gunlink() and H5Grename(). ./src/H5F.c ./src/H5Fistore.c ./src/H5Fprivate.h Removed the last memcpy() from the chunk cache. ./src/H5Fistore.c The offset of a chunk within a dataset is an 8-byte quantity per dimension instead of 4 bytes. ./src/H5HL.c Fixed infinite loops in H5HL_remove().
1998-09-28 22:20:21 +08:00
{
[svn-r731] Changes since 19980930 ---------------------- ./src/H5G.c ./test/unlink.c Renaming a symbolic link works like on Unix: the link is renamed but the link value is unchanged.
1998-10-01 04:17:26 +08:00
H5G_stat_t sb;
char *linkval=NULL;
size_t lv_size=32;
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
H5G_entry_t obj_ent; /* Object entry for object being moved */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
H5RS_str_t *src_name_r; /* Ref-counted version of src name */
H5RS_str_t *dst_name_r; /* Ref-counted version of dest name */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r731] Changes since 19980930 ---------------------- ./src/H5G.c ./test/unlink.c Renaming a symbolic link works like on Unix: the link is renamed but the link value is unchanged.
1998-10-01 04:17:26 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_move, FAIL);
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
assert(src_loc);
assert(dst_loc);
[svn-r726] Changes since 19980924 ---------------------- ./MANIFEST ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./test/Makefile.in ./test/unlink.c [NEW] Finished H5Gunlink() and H5Grename(). ./src/H5F.c ./src/H5Fistore.c ./src/H5Fprivate.h Removed the last memcpy() from the chunk cache. ./src/H5Fistore.c The offset of a chunk within a dataset is an 8-byte quantity per dimension instead of 4 bytes. ./src/H5HL.c Fixed infinite loops in H5HL_remove().
1998-09-28 22:20:21 +08:00
assert(src_name && *src_name);
assert(dst_name && *dst_name);
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_get_objinfo(src_loc, src_name, FALSE, &sb, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, FAIL, "object not found");
[svn-r731] Changes since 19980930 ---------------------- ./src/H5G.c ./test/unlink.c Renaming a symbolic link works like on Unix: the link is renamed but the link value is unchanged.
1998-10-01 04:17:26 +08:00
if (H5G_LINK==sb.type) {
/*
* When renaming a symbolic link we rename the link but don't change
* the value of the link.
*/
do {
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
if (NULL==(linkval=H5MM_realloc(linkval, 2*lv_size)))
HGOTO_ERROR(H5E_RESOURCE, H5E_NOSPACE, FAIL, "unable to allocate space for symbolic link value");
[svn-r731] Changes since 19980930 ---------------------- ./src/H5G.c ./test/unlink.c Renaming a symbolic link works like on Unix: the link is renamed but the link value is unchanged.
1998-10-01 04:17:26 +08:00
linkval[lv_size-1] = '\0';
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_linkval(src_loc, src_name, lv_size, linkval, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, FAIL, "unable to read symbolic link value");
[svn-r731] Changes since 19980930 ---------------------- ./src/H5G.c ./test/unlink.c Renaming a symbolic link works like on Unix: the link is renamed but the link value is unchanged.
1998-10-01 04:17:26 +08:00
} while (linkval[lv_size-1]);
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
if (H5G_link(src_loc, linkval, dst_loc, dst_name, H5G_LINK_SOFT,
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
H5G_TARGET_NORMAL, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, FAIL, "unable to rename symbolic link");
[svn-r731] Changes since 19980930 ---------------------- ./src/H5G.c ./test/unlink.c Renaming a symbolic link works like on Unix: the link is renamed but the link value is unchanged.
1998-10-01 04:17:26 +08:00
H5MM_xfree(linkval);
} else {
/*
* Rename the object.
*/
[svn-r5273] Purpose: New feature Description: Allow H5Glink and H5Gmove to handle links across different locations. Solution: Added H5Glink2 and H5Gmove2 functions with new parameter of destination location. Platforms tested: Linux 2.2(eirene)
2002-04-27 04:34:46 +08:00
if (H5G_link(src_loc, src_name, dst_loc, dst_name, H5G_LINK_HARD,
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
H5G_TARGET_MOUNT, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, FAIL, "unable to register new name for object");
[svn-r731] Changes since 19980930 ---------------------- ./src/H5G.c ./test/unlink.c Renaming a symbolic link works like on Unix: the link is renamed but the link value is unchanged.
1998-10-01 04:17:26 +08:00
}
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Search the open ID list and replace names for the move operation
* This has to be done here because H5G_link and H5G_unlink have
* internal object entries, and do not modify the entries list
*/
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_namei(src_loc, src_name, NULL, NULL, &obj_ent, H5G_TARGET_NORMAL|H5G_TARGET_SLINK, NULL, H5G_NAMEI_TRAVERSE, NULL, dxpl_id))
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
HGOTO_ERROR (H5E_SYM, H5E_NOTFOUND, FAIL, "unable to follow symbolic link");
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
src_name_r=H5RS_wrap(src_name);
assert(src_name_r);
dst_name_r=H5RS_wrap(dst_name);
assert(dst_name_r);
if (H5G_replace_name(sb.type, &obj_ent, src_name_r, src_loc, dst_name_r, dst_loc, OP_MOVE )<0)
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, FAIL, "unable to replace name ");
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
H5RS_decr(src_name_r);
H5RS_decr(dst_name_r);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
H5G_free_ent_name(&obj_ent);
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
[svn-r731] Changes since 19980930 ---------------------- ./src/H5G.c ./test/unlink.c Renaming a symbolic link works like on Unix: the link is renamed but the link value is unchanged.
1998-10-01 04:17:26 +08:00
/* Remove the old name */
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_unlink(src_loc, src_name, dxpl_id)<0)
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
HGOTO_ERROR(H5E_SYM, H5E_CANTINIT, FAIL, "unable to deregister old object name");
[svn-r726] Changes since 19980924 ---------------------- ./MANIFEST ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./test/Makefile.in ./test/unlink.c [NEW] Finished H5Gunlink() and H5Grename(). ./src/H5F.c ./src/H5Fistore.c ./src/H5Fprivate.h Removed the last memcpy() from the chunk cache. ./src/H5Fistore.c The offset of a chunk within a dataset is an 8-byte quantity per dimension instead of 4 bytes. ./src/H5HL.c Fixed infinite loops in H5HL_remove().
1998-09-28 22:20:21 +08:00
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r726] Changes since 19980924 ---------------------- ./MANIFEST ./src/H5B.c ./src/H5Bprivate.h ./src/H5G.c ./src/H5Gnode.c ./src/H5Gprivate.h ./test/Makefile.in ./test/unlink.c [NEW] Finished H5Gunlink() and H5Grename(). ./src/H5F.c ./src/H5Fistore.c ./src/H5Fprivate.h Removed the last memcpy() from the chunk cache. ./src/H5Fistore.c The offset of a chunk within a dataset is an 8-byte quantity per dimension instead of 4 bytes. ./src/H5HL.c Fixed infinite loops in H5HL_remove().
1998-09-28 22:20:21 +08:00
}
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_insertion_file
*
* Purpose: Given a location and name that specifies a not-yet-existing
* object return the file into which the object is about to be
* inserted.
*
* Return: Success: File pointer
*
* Failure: NULL
*
* Programmer: Robb Matzke
* Wednesday, October 14, 1998
*
* Modifications:
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Pedro Vicente, <pvn@ncsa.uiuc.edu> 18 Sep 2002
* Added `id to name' support.
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
*
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
*-------------------------------------------------------------------------
*/
H5F_t *
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
H5G_insertion_file(H5G_entry_t *loc, const char *name, hid_t dxpl_id)
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
{
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
H5F_t *ret_value; /* Return value */
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
[svn-r5471] Purpose: Code cleanup Description: Broke the FUNC_ENTER macro into several macros, with more specialized uses (which followup mail will describe). This was designed to move most/all of the checks which could be done at compile time to that point, instead of needlessly performing them (over & over :-) at run-time. This reduces the library's size (and thus staticly linked binaries) and has a minor speedup effect also. Platforms tested: IRIX64 6.5 (modi4) with parallel & FORTRAN enabled, and additional testing on FreeBSD and Solaris immediately after the checkin.
2002-05-29 23:07:55 +08:00
FUNC_ENTER_NOAPI(H5G_insertion_file, NULL);
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
assert(loc);
assert(name && *name);
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/* Check if the location the object will be inserted into is part of a
* file mounting chain (either a parent or a child) and perform a more
* rigorous determination of the location's file (which traverses into
* mounted files, etc.).
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
*/
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(H5F_has_mount(loc->file) || H5F_is_mount(loc->file)) {
const char *rest;
H5G_entry_t grp_ent;
size_t size;
/*
* Look up the name to get the containing group and to make sure the name
* doesn't already exist.
*/
[svn-r6387] Purpose: Bug Fix Description: Metadata cache in parallel I/O can cause hangs in applications which perform independent I/O on chunked datasets, because the metadata cache can attempt to flush out dirty metadata from only a single process, instead of collectively from all processes. Solution: Pass a dataset transfer property list down from every API function which could possibly trigger metadata I/O. Then, split the metadata cache into two sets of entries to allow dirty metadata to be set aside when a hash table collision occurs during independent I/O. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} FreeBSD 4.7 (sleipnir) serial & parallel Misc. update: Updated release_docs/RELEASE
2003-02-11 01:26:09 +08:00
if (H5G_namei(loc, name, &rest, &grp_ent, NULL, H5G_TARGET_NORMAL, NULL, H5G_NAMEI_TRAVERSE, NULL, dxpl_id)>=0) {
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
H5G_free_ent_name(&grp_ent);
HGOTO_ERROR(H5E_SYM, H5E_EXISTS, NULL, "name already exists");
} /* end if */
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
H5E_clear(NULL);
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/* Make sure only the last component wasn't resolved */
rest = H5G_component(rest, &size);
assert(*rest && size>0);
rest = H5G_component(rest+size, NULL);
if (*rest) {
H5G_free_ent_name(&grp_ent);
HGOTO_ERROR(H5E_SYM, H5E_NOTFOUND, NULL, "insertion point not found");
} /* end if */
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/* Set return value */
ret_value=grp_ent.file;
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/* Free the ID to name buffer */
H5G_free_ent_name(&grp_ent);
} /* end if */
else
/* Use the location's file */
ret_value=loc->file;
[svn-r5842] Purpose: Code cleanup Description: Change most (all?) HRETURN_ERROR macros to HGOTO_ERROR macros, along with HRETURN macros to HGOTO_DONE macros. This unifies the error return path from functions and reduces the size of the library by up to 10% on some platforms. Additionally, I improved a lot of the error cleanup code in many routines. Platforms tested: FreeBSD 4.6 (sleipnir) serial & parallel and IRIX64 6.5 (modi4) serial & parallel.
2002-08-09 00:52:55 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r761] Changes since 19981013 ---------------------- ./src/H5.c Fixed a signed vs. unsigned comparison. ./src/H5D.c Setting a fill value of all zeros will cause the fill value to be written to the dataset instead of relying on the low-level file driver initializing unwritten areas with zero. ./src/H5D.c ./src/H5F.c ./src/H5Fprivate.h ./src/H5G.c ./src/H5Gpkg.h ./src/H5Gprivate.h ./src/H5O.c ./src/H5T.c More file mounting stuff. ./src/H5I.c Fixed a bug where trying to close an invalid object id caused a core dump. For instance, H5Gclose(-1). ./MANIFEST ./test/Makefile.in ./test/mount.c [NEW] Mounting tests. ./src/H5R.c Fixed a couple (herr_t)NULL casts.
1998-10-15 03:35:08 +08:00
}
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_free_grp_name
*
* Purpose: Free the 'ID to name' buffers.
*
* Return: Non-negative on success/Negative on failure
*
* Programmer: Pedro Vicente, pvn@ncsa.uiuc.edu
*
* Date: August 22, 2002
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Comments: Used now only on the root group close, in H5F_close()
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
herr_t
H5G_free_grp_name(H5G_t *grp)
{
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
H5G_entry_t *ent; /* Group object's entry */
herr_t ret_value=SUCCEED; /* Return value */
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
FUNC_ENTER_NOAPI(H5G_free_grp_name, FAIL);
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Check args */
assert(grp);
assert(grp->nref > 0);
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Get the entry for the group */
if (NULL==( ent = H5G_entof(grp)))
HGOTO_ERROR (H5E_SYM, H5E_CANTINIT, FAIL, "cannot get entry");
/* Free the entry */
H5G_free_ent_name(ent);
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
}
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_free_ent_name
*
* Purpose: Free the 'ID to name' buffers.
*
* Return: Success
*
* Programmer: Pedro Vicente, pvn@ncsa.uiuc.edu
*
* Date: August 22, 2002
*
* Comments:
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
herr_t
H5G_free_ent_name(H5G_entry_t *ent)
{
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
herr_t ret_value=SUCCEED; /* Return value */
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
FUNC_ENTER_NOAPI(H5G_free_ent_name, FAIL);
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Check args */
assert(ent);
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(ent->user_path_r) {
H5RS_decr(ent->user_path_r);
ent->user_path_r=NULL;
} /* end if */
if(ent->canon_path_r) {
H5RS_decr(ent->canon_path_r);
ent->canon_path_r=NULL;
} /* end if */
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
}
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_replace_name
*
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
* Purpose: Search the list of open IDs and replace names according to a
* particular operation. The operation occured on the LOC
* entry, which had SRC_NAME previously. The new name (if there
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
* is one) is DST_NAME. Additional entry location information
* (currently only needed for the 'move' operation) is passed
* in SRC_LOC and DST_LOC.
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
*
* Return: Success: 0, Failure: -1
*
* Programmer: Pedro Vicente, pvn@ncsa.uiuc.edu
*
* Date: June 11, 2002
*
* Comments:
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
herr_t
[svn-r6779] Purpose: Backward compatibility change Description: 1.4 compatibility for H5G_obj_t type and H5Zregister test. Solution: use macro H5_WANT_H5_V1_4_COMPAT Platforms tested: h5committest
2003-04-30 03:49:48 +08:00
H5G_replace_name(int type, H5G_entry_t *loc,
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
H5RS_str_t *src_name, H5G_entry_t *src_loc,
H5RS_str_t *dst_name, H5G_entry_t *dst_loc, H5G_names_op_t op )
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
{
H5G_names_t names; /* Structure to hold operation information for callback */
unsigned search_group=0; /* Flag to indicate that groups are to be searched */
unsigned search_dataset=0; /* Flag to indicate that datasets are to be searched */
unsigned search_datatype=0; /* Flag to indicate that datatypes are to be searched */
herr_t ret_value = SUCCEED;
FUNC_ENTER_NOAPI(H5G_replace_name, FAIL);
/* Set up common information for callback */
names.src_name=src_name;
names.dst_name=dst_name;
names.loc=loc;
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
names.src_loc=src_loc;
names.dst_loc=dst_loc;
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
names.op=op;
/* Determine which types of IDs need to be operated on */
switch(type) {
/* Object is a group */
case H5G_GROUP:
/* Search and replace names through group IDs */
search_group=1;
break;
/* Object is a dataset */
case H5G_DATASET:
/* Search and replace names through dataset IDs */
search_dataset=1;
break;
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
[svn-r7313] Purpose: Code cleanup & feature change Description: Chase error API changes Also, switch from using 'unsigned long[2]' for the objno in the H5G_stat_t struct to just using 'haddr_t', since 'haddr_t' is already part of the public information for the library and the 'unsigned long[2]' form was just confusing and obscured the issue. Platforms tested: h5committested
2003-08-09 03:12:58 +08:00
/* Object is a named datatype */
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
case H5G_TYPE:
/* Search and replace names through datatype IDs */
search_datatype=1;
break;
[svn-r5911] Purpose: bug fix in 'ID to name' function Description: the function replace_name was only checking for immutable datatypes Solution: added a new function H5T_is_named, that checks for named datatypes Platforms tested: windows 2000, linux, solaris with Fortran
2002-09-04 23:50:12 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
case H5G_UNKNOWN: /* We pass H5G_UNKNOWN as object type when we need to search all IDs */
case H5G_LINK: /* Symbolic links might resolve to any object, so we need to search all IDs */
/* Check if we will need to search groups */
if(H5I_nmembers(H5I_GROUP)>0)
search_group=1;
[svn-r5911] Purpose: bug fix in 'ID to name' function Description: the function replace_name was only checking for immutable datatypes Solution: added a new function H5T_is_named, that checks for named datatypes Platforms tested: windows 2000, linux, solaris with Fortran
2002-09-04 23:50:12 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Check if we will need to search datasets */
if(H5I_nmembers(H5I_DATASET)>0)
search_dataset=1;
/* Check if we will need to search datatypes */
if(H5I_nmembers(H5I_DATATYPE)>0)
search_datatype=1;
break;
default:
HGOTO_ERROR (H5E_DATATYPE, H5E_BADTYPE, FAIL, "not valid object type");
} /* end switch */
/* Search through group IDs */
if(search_group)
H5I_search(H5I_GROUP, H5G_replace_ent, &names);
/* Search through dataset IDs */
if(search_dataset)
H5I_search(H5I_DATASET, H5G_replace_ent, &names);
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Search through datatype IDs */
if(search_datatype)
H5I_search(H5I_DATATYPE, H5G_replace_ent, &names);
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
}
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_common_path
*
* Purpose: Determine if one path is a valid prefix of another path
*
* Return: TRUE for valid prefix, FALSE for not a valid prefix, FAIL
* on error
*
* Programmer: Quincey Koziol, koziol@ncsa.uiuc.edu
*
* Date: September 24, 2002
*
* Comments:
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
static htri_t
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
H5G_common_path(const H5RS_str_t *fullpath_r, const H5RS_str_t *prefix_r)
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
{
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
const char *fullpath; /* Pointer to actual fullpath string */
const char *prefix; /* Pointer to actual prefix string */
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
size_t nchars1,nchars2; /* Number of characters in components */
htri_t ret_value=FALSE; /* Return value */
FUNC_ENTER_NOINIT(H5G_common_path);
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Get component of each name */
[svn-r6433] Purpose: Code/comment cleanup Description: Made more information about reference-counted strings private to the H5RS package by trimming the typedef used in H5RSprivate.h and switching from using macros for a couple of operations to using functions. Updated copyright information for a few of these files. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)}
2003-02-25 04:19:01 +08:00
fullpath=H5RS_get_str(fullpath_r);
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
assert(fullpath);
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
fullpath=H5G_component(fullpath,&nchars1);
assert(fullpath);
[svn-r6433] Purpose: Code/comment cleanup Description: Made more information about reference-counted strings private to the H5RS package by trimming the typedef used in H5RSprivate.h and switching from using macros for a couple of operations to using functions. Updated copyright information for a few of these files. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)}
2003-02-25 04:19:01 +08:00
prefix=H5RS_get_str(prefix_r);
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
assert(prefix);
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
prefix=H5G_component(prefix,&nchars2);
assert(prefix);
/* Check if we have a real string for each component */
while(*fullpath && *prefix) {
/* Check that the components we found are the same length */
if(nchars1==nchars2) {
/* Check that the two components are equal */
if(HDstrncmp(fullpath,prefix,nchars1)==0) {
/* Advance the pointers in the names */
fullpath+=nchars1;
prefix+=nchars2;
/* Get next component of each name */
fullpath=H5G_component(fullpath,&nchars1);
assert(fullpath);
prefix=H5G_component(prefix,&nchars2);
assert(prefix);
} /* end if */
else
HGOTO_DONE(FALSE);
} /* end if */
else
HGOTO_DONE(FALSE);
} /* end while */
/* If we reached the end of the prefix path to check, it must be a valid prefix */
if(*prefix=='\0')
ret_value=TRUE;
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
}
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
/*-------------------------------------------------------------------------
* Function: H5G_replace_ent
*
* Purpose: H5I_search callback function to replace group entry names
*
* Return: Success: 0, Failure: -1
*
* Programmer: Pedro Vicente, pvn@ncsa.uiuc.edu
*
* Date: June 5, 2002
*
* Comments:
*
* Modifications:
*
*-------------------------------------------------------------------------
*/
[svn-r7071] Purpose: Code cleanup Description: Tweak internal iterator over IDs to not require constant "operator" data. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest
2003-06-20 23:19:18 +08:00
static int
H5G_replace_ent(void *obj_ptr, hid_t obj_id, void *key)
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
{
const H5G_names_t *names = (const H5G_names_t *)key; /* Get operation's information */
H5G_entry_t *ent = NULL; /* Group entry for object that the ID refers to */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
H5F_t *top_ent_file; /* Top file in entry's mounted file chain */
H5F_t *top_loc_file; /* Top file in location's mounted file chain */
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
herr_t ret_value = SUCCEED; /* Return value */
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
FUNC_ENTER_NOINIT(H5G_replace_ent);
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
assert(obj_ptr);
[svn-r5911] Purpose: bug fix in 'ID to name' function Description: the function replace_name was only checking for immutable datatypes Solution: added a new function H5T_is_named, that checks for named datatypes Platforms tested: windows 2000, linux, solaris with Fortran
2002-09-04 23:50:12 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
/* Get the symbol table entry */
[svn-r6878] Purpose: Code cleanup Description: Limit the scope on more function prototypes/macros/typedefs. Platforms tested: FreeBSD 4.8 (sleipnir) h5committest not necessary.
2003-05-16 03:22:33 +08:00
switch(H5I_get_type(obj_id)) {
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
case H5I_GROUP:
ent = H5G_entof((H5G_t*)obj_ptr);
break;
case H5I_DATASET:
ent = H5D_entof((H5D_t*)obj_ptr);
break;
case H5I_DATATYPE:
/* Avoid non-named datatypes */
if(!H5T_is_named((H5T_t*)obj_ptr))
HGOTO_DONE(SUCCEED); /* Do not exit search over IDs */
[svn-r5911] Purpose: bug fix in 'ID to name' function Description: the function replace_name was only checking for immutable datatypes Solution: added a new function H5T_is_named, that checks for named datatypes Platforms tested: windows 2000, linux, solaris with Fortran
2002-09-04 23:50:12 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
ent = H5T_entof((H5T_t*)obj_ptr);
break;
[svn-r5929] Purpose: bug fix Description: memory leak regarding the ID to name buffer Solution: added a new function H5G_free_ent_name that is called on several places of the library Platforms tested: windows 2000 linux, with cpp solaris, with fortran, cpp irix64, with parallel, fortran
2002-09-18 23:51:29 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
default:
HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, FAIL, "unknown data object");
} /* end switch */
assert(ent);
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
switch(names->op) {
/*-------------------------------------------------------------------------
* OP_MOUNT
*-------------------------------------------------------------------------
*/
case OP_MOUNT:
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(ent->file->mtab.parent && H5RS_cmp(ent->user_path_r,ent->canon_path_r)) {
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Find the "top" file in the chain of mounted files */
top_ent_file=ent->file->mtab.parent;
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
while(top_ent_file->mtab.parent!=NULL)
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
top_ent_file=top_ent_file->mtab.parent;
} /* end if */
else
top_ent_file=ent->file;
/* Check for entry being in correct file (or mounted file) */
if(top_ent_file->shared == names->loc->file->shared) {
/* Check if the source is along the entry's path */
/* (But not actually the entry itself) */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(H5G_common_path(ent->user_path_r,names->src_name) &&
H5RS_cmp(ent->user_path_r,names->src_name)!=0) {
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Hide the user path */
ent->user_path_hidden++;
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
} /* end if */
} /* end if */
break;
/*-------------------------------------------------------------------------
* OP_UNMOUNT
*-------------------------------------------------------------------------
*/
case OP_UNMOUNT:
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
if(ent->file->mtab.parent) {
/* Find the "top" file in the chain of mounted files for the entry */
top_ent_file=ent->file->mtab.parent;
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
while(top_ent_file->mtab.parent!=NULL)
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
top_ent_file=top_ent_file->mtab.parent;
} /* end if */
else
top_ent_file=ent->file;
if(names->loc->file->mtab.parent) {
/* Find the "top" file in the chain of mounted files for the location */
top_loc_file=names->loc->file->mtab.parent;
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
while(top_loc_file->mtab.parent!=NULL)
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
top_loc_file=top_loc_file->mtab.parent;
} /* end if */
else
top_loc_file=names->loc->file;
if(ent->user_path_hidden) {
/* If the ID's entry is not in the file we operated on, skip it */
if(top_ent_file->shared == top_loc_file->shared) {
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(H5G_common_path(ent->user_path_r,names->src_name)) {
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Un-hide the user path */
ent->user_path_hidden--;
} /* end if */
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
} /* end if */
} /* end if */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
else {
/* If the ID's entry is not in the file we operated on, skip it */
if(top_ent_file->shared == top_loc_file->shared) {
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(ent->user_path_r && H5G_common_path(ent->user_path_r,names->src_name)) {
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Free user path */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(ent->user_path_r!=NULL) {
H5RS_decr(ent->user_path_r);
ent->user_path_r=NULL;
} /* end if */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
} /* end if */
} /* end if */
} /* end else */
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
break;
/*-------------------------------------------------------------------------
* OP_UNLINK
*-------------------------------------------------------------------------
*/
case OP_UNLINK:
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* If the ID's entry is not in the file we operated on, skip it */
if(ent->file->shared == names->loc->file->shared) {
/* Check if we are referring to the same object */
if(H5F_addr_eq(ent->header, names->loc->header)) {
/* Check if the object was opened with the same canonical path as the one being moved */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(H5RS_cmp(ent->canon_path_r,names->loc->canon_path_r)==0) {
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Free user path */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(ent->user_path_r!=NULL) {
H5RS_decr(ent->user_path_r);
ent->user_path_r=NULL;
} /* end if */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
} /* end if */
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
} /* end if */
else {
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Check if the location being unlinked is in the canonical path for the current object */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(H5G_common_path(ent->canon_path_r,names->loc->canon_path_r)) {
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Free user path */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(ent->user_path_r!=NULL) {
H5RS_decr(ent->user_path_r);
ent->user_path_r=NULL;
} /* end if */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
} /* end if */
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
} /* end else */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
} /* end if */
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
break;
/*-------------------------------------------------------------------------
* OP_MOVE
*-------------------------------------------------------------------------
*/
case OP_MOVE: /* H5Gmove case, check for relative names case */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* If the ID's entry is not in the file we operated on, skip it */
if(ent->file->shared == names->loc->file->shared) {
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
H5RS_str_t *src_path_r; /* Full user path of source name */
H5RS_str_t *dst_path_r; /* Full user path of destination name */
H5RS_str_t *canon_src_path_r; /* Copy of canonical part of source path */
H5RS_str_t *canon_dst_path_r; /* Copy of canonical part of destination path */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Make certain that the source and destination names are full (not relative) paths */
[svn-r6433] Purpose: Code/comment cleanup Description: Made more information about reference-counted strings private to the H5RS package by trimming the typedef used in H5RSprivate.h and switching from using macros for a couple of operations to using functions. Updated copyright information for a few of these files. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)}
2003-02-25 04:19:01 +08:00
if(*(H5RS_get_str(names->src_name))!='/') {
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
const char *src_name; /* Pointer to raw string of src_name */
char *src_path; /* Full user path of source name */
const char *src_user_path; /* Pointer to raw string of src path */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
size_t src_path_len; /* Length of the source path */
unsigned need_sep; /* Flag to indicate if separator is needed */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/* Get the pointer to the raw src user path */
[svn-r6433] Purpose: Code/comment cleanup Description: Made more information about reference-counted strings private to the H5RS package by trimming the typedef used in H5RSprivate.h and switching from using macros for a couple of operations to using functions. Updated copyright information for a few of these files. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)}
2003-02-25 04:19:01 +08:00
src_user_path=H5RS_get_str(names->src_loc->user_path_r);
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Get the length of the name for the source group's user path */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
src_path_len=HDstrlen(src_user_path);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Determine if there is a trailing separator in the name */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(src_user_path[src_path_len-1]=='/')
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
need_sep=0;
else
need_sep=1;
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/* Get the pointer to the raw src user path */
[svn-r6433] Purpose: Code/comment cleanup Description: Made more information about reference-counted strings private to the H5RS package by trimming the typedef used in H5RSprivate.h and switching from using macros for a couple of operations to using functions. Updated copyright information for a few of these files. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)}
2003-02-25 04:19:01 +08:00
src_name=H5RS_get_str(names->src_name);
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Add in the length needed for the '/' separator and the relative path */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
src_path_len+=HDstrlen(src_name)+need_sep;
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Allocate space for the path */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(NULL==(src_path = H5FL_BLK_MALLOC(str_buf,src_path_len+1)))
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
HGOTO_ERROR (H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed");
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
HDstrcpy(src_path,src_user_path);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
if(need_sep)
HDstrcat(src_path,"/");
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
HDstrcat(src_path,src_name);
/* Create reference counted string for src path */
src_path_r=H5RS_own(src_path);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
} /* end if */
else
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
src_path_r=H5RS_dup(names->src_name);
[svn-r6433] Purpose: Code/comment cleanup Description: Made more information about reference-counted strings private to the H5RS package by trimming the typedef used in H5RSprivate.h and switching from using macros for a couple of operations to using functions. Updated copyright information for a few of these files. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)}
2003-02-25 04:19:01 +08:00
if(*(H5RS_get_str(names->dst_name))!='/') {
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
const char *dst_name; /* Pointer to raw string of dst_name */
char *dst_path; /* Full user path of destination name */
const char *dst_user_path; /* Pointer to raw string of dst path */
size_t dst_path_len; /* Length of the destination path */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
unsigned need_sep; /* Flag to indicate if separator is needed */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/* Get the pointer to the raw dst user path */
[svn-r6433] Purpose: Code/comment cleanup Description: Made more information about reference-counted strings private to the H5RS package by trimming the typedef used in H5RSprivate.h and switching from using macros for a couple of operations to using functions. Updated copyright information for a few of these files. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)}
2003-02-25 04:19:01 +08:00
dst_user_path=H5RS_get_str(names->dst_loc->user_path_r);
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Get the length of the name for the destination group's user path */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
dst_path_len=HDstrlen(dst_user_path);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Determine if there is a trailing separator in the name */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(dst_user_path[dst_path_len-1]=='/')
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
need_sep=0;
else
need_sep=1;
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/* Get the pointer to the raw dst user path */
[svn-r6433] Purpose: Code/comment cleanup Description: Made more information about reference-counted strings private to the H5RS package by trimming the typedef used in H5RSprivate.h and switching from using macros for a couple of operations to using functions. Updated copyright information for a few of these files. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)}
2003-02-25 04:19:01 +08:00
dst_name=H5RS_get_str(names->dst_name);
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Add in the length needed for the '/' separator and the relative path */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
dst_path_len+=HDstrlen(dst_name)+need_sep;
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Allocate space for the path */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(NULL==(dst_path = H5FL_BLK_MALLOC(str_buf,dst_path_len+1)))
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
HGOTO_ERROR (H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed");
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
HDstrcpy(dst_path,dst_user_path);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
if(need_sep)
HDstrcat(dst_path,"/");
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
HDstrcat(dst_path,dst_name);
/* Create reference counted string for dst path */
dst_path_r=H5RS_own(dst_path);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
} /* end if */
else
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
dst_path_r=H5RS_dup(names->dst_name);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Get the canonical parts of the source and destination names */
/* Check if the object being moved was accessed through a mounted file */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(H5RS_cmp(names->loc->user_path_r,names->loc->canon_path_r)!=0) {
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
size_t non_canon_name_len; /* Length of non-canonical part of name */
/* Get current string lengths */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
non_canon_name_len=H5RS_len(names->loc->user_path_r)-H5RS_len(names->loc->canon_path_r);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
[svn-r6433] Purpose: Code/comment cleanup Description: Made more information about reference-counted strings private to the H5RS package by trimming the typedef used in H5RSprivate.h and switching from using macros for a couple of operations to using functions. Updated copyright information for a few of these files. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)}
2003-02-25 04:19:01 +08:00
canon_src_path_r=H5RS_create(H5RS_get_str(src_path_r)+non_canon_name_len);
canon_dst_path_r=H5RS_create(H5RS_get_str(dst_path_r)+non_canon_name_len);
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
} /* end if */
else {
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
canon_src_path_r=H5RS_dup(src_path_r);
canon_dst_path_r=H5RS_dup(dst_path_r);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
} /* end else */
/* Check if the link being changed in the file is along the canonical path for this object */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(H5G_common_path(ent->canon_path_r,canon_src_path_r)) {
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
size_t user_dst_len; /* Length of destination user path */
size_t canon_dst_len; /* Length of destination canonical path */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
const char *old_user_path; /* Pointer to previous user path */
char *new_user_path; /* Pointer to new user path */
char *new_canon_path; /* Pointer to new canonical path */
const char *tail_path; /* Pointer to "tail" of path */
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
size_t tail_len; /* Pointer to "tail" of path */
char *src_canon_prefix; /* Pointer to source canonical path prefix of component which is moving */
size_t src_canon_prefix_len;/* Length of the source canonical path prefix */
char *dst_canon_prefix; /* Pointer to destination canonical path prefix of component which is moving */
size_t dst_canon_prefix_len;/* Length of the destination canonical path prefix */
char *user_prefix; /* Pointer to user path prefix of component which is moving */
size_t user_prefix_len; /* Length of the user path prefix */
char *src_comp; /* The source name of the component which is actually changing */
char *dst_comp; /* The destination name of the component which is actually changing */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
const char *canon_src_path; /* pointer to canonical part of source path */
const char *canon_dst_path; /* pointer to canonical part of destination path */
/* Get the pointers to the raw strings */
[svn-r6433] Purpose: Code/comment cleanup Description: Made more information about reference-counted strings private to the H5RS package by trimming the typedef used in H5RSprivate.h and switching from using macros for a couple of operations to using functions. Updated copyright information for a few of these files. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)}
2003-02-25 04:19:01 +08:00
canon_src_path=H5RS_get_str(canon_src_path_r);
canon_dst_path=H5RS_get_str(canon_dst_path_r);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Get the source & destination components */
src_comp=HDstrrchr(canon_src_path,'/');
assert(src_comp);
dst_comp=HDstrrchr(canon_dst_path,'/');
assert(dst_comp);
/* Find the canonical prefixes for the entry */
src_canon_prefix_len=HDstrlen(canon_src_path)-HDstrlen(src_comp);
if(NULL==(src_canon_prefix = H5MM_malloc(src_canon_prefix_len+1)))
HGOTO_ERROR (H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed");
HDstrncpy(src_canon_prefix,canon_src_path,src_canon_prefix_len);
src_canon_prefix[src_canon_prefix_len]='\0';
dst_canon_prefix_len=HDstrlen(canon_dst_path)-HDstrlen(dst_comp);
if(NULL==(dst_canon_prefix = H5MM_malloc(dst_canon_prefix_len+1)))
HGOTO_ERROR (H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed");
HDstrncpy(dst_canon_prefix,canon_dst_path,dst_canon_prefix_len);
dst_canon_prefix[dst_canon_prefix_len]='\0';
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
/* Hold this for later use */
[svn-r6433] Purpose: Code/comment cleanup Description: Made more information about reference-counted strings private to the H5RS package by trimming the typedef used in H5RSprivate.h and switching from using macros for a couple of operations to using functions. Updated copyright information for a few of these files. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)}
2003-02-25 04:19:01 +08:00
old_user_path=H5RS_get_str(ent->user_path_r);
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Find the user prefix for the entry */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
user_prefix_len=HDstrlen(old_user_path)-H5RS_len(ent->canon_path_r);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
if(NULL==(user_prefix = H5MM_malloc(user_prefix_len+1)))
HGOTO_ERROR (H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed");
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
HDstrncpy(user_prefix,old_user_path,user_prefix_len);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
user_prefix[user_prefix_len]='\0';
/* Set the tail path info */
tail_path=old_user_path+user_prefix_len+src_canon_prefix_len+HDstrlen(src_comp);
tail_len=HDstrlen(tail_path);
/* Get the length of the destination paths */
user_dst_len=user_prefix_len+dst_canon_prefix_len+HDstrlen(dst_comp)+tail_len;
canon_dst_len=dst_canon_prefix_len+HDstrlen(dst_comp)+tail_len;
/* Allocate space for the new user path */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(NULL==(new_user_path = H5FL_BLK_MALLOC(str_buf,user_dst_len+1)))
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
HGOTO_ERROR (H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed");
/* Allocate space for the new canonical path */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
if(NULL==(new_canon_path = H5FL_BLK_MALLOC(str_buf,canon_dst_len+1)))
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
HGOTO_ERROR (H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed");
/* Create the new names */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
HDstrcpy(new_user_path,user_prefix);
HDstrcat(new_user_path,dst_canon_prefix);
HDstrcat(new_user_path,dst_comp);
HDstrcat(new_user_path,tail_path);
HDstrcpy(new_canon_path,dst_canon_prefix);
HDstrcat(new_canon_path,dst_comp);
HDstrcat(new_canon_path,tail_path);
/* Release the old user & canonical paths */
H5RS_decr(ent->user_path_r);
H5RS_decr(ent->canon_path_r);
/* Take ownership of the new user & canonical paths */
ent->user_path_r=H5RS_own(new_user_path);
ent->canon_path_r=H5RS_own(new_canon_path);
[svn-r5995] Purpose: Code cleanup/Bug Fix Description: Re-do the ID->name code again, taking into account many more weird and wonderful special cases stumbled across during the last set of test writing. Platforms tested: FreeBSD 4.6 (sleipnir) w and w/o parallel Linux 2.2.x (eirene) w/FORTRAN & C++ Solaris 2.7 (arabica) w/FORTRAN IRIX64 6.5 (modi4) w/FORTRAN & parallel
2002-10-15 04:08:23 +08:00
/* Free the extra paths allocated */
H5MM_xfree(src_canon_prefix);
H5MM_xfree(dst_canon_prefix);
H5MM_xfree(user_prefix);
} /* end if */
/* Free the extra paths allocated */
[svn-r6252] Purpose: Lots of performance improvements & a couple new internal API interfaces. Description: Performance Improvements: - Cached file offset & length sizes in shared file struct, to avoid constantly looking them up in the FCPL. - Generic property improvements: - Added "revision" number to generic property classes to speed up comparisons. - Changed method of storing properties from using a hash-table to the TBBT routines in the library. - Share the propery names between classes and the lists derived from them. - Removed redundant 'def_value' buffer from each property. - Switching code to use a "copy on write" strategy for properties in each list, where the properties in each list are shared with the properties in the class, until a property's value is changed in a list. - Fixed error in layout code which was allocating too many buffers. - Redefined public macros of the form (H5open()/H5check, <variable>) internally to only be (<variable>), avoiding innumerable useless calls to H5open() and H5check_version(). - Reuse already zeroed buffers in H5F_contig_fill instead of constantly re-zeroing them. - Don't write fill values if writing entire dataset. - Use gettimeofday() system call instead of time() system when checking the modification time of a dataset. - Added reference counted string API and use it for tracking the names of objects opening in a file (for the ID->name code). - Removed redundant H5P_get() calls in B-tree routines. - Redefine H5T datatype macros internally to the library, to avoid calling H5check redundantly. - Keep dataspace information for dataset locally instead of reading from disk each time. Added new module to track open objects in a file, to allow this (which will be useful eventually for some FPH5 metadata caching issues). - Remove H5AC_find macro which was inlining metadata cache lookups, and call function instead. - Remove redundant memset() calls from H5G_namei() routine. - Remove redundant checking of object type when locating objects in metadata cache and rely on the address only. - Create default dataset object to use when default dataset creation property list is used to create datasets, bypassing querying for all the property list values. - Use default I/O vector size when performing raw data with the default dataset transfer property list, instead of querying for I/O vector size. - Remove H5P_DEFAULT internally to the library, replacing it with more specific default property list based on the type of property list needed. - Remove redundant memset() calls in object header message (H5O*) routines. - Remove redunant memset() calls in data I/O routines. - Split free-list allocation routines into malloc() and calloc()- like routines, instead of one combined routine. - Remove lots of indirection in H5O*() routines. - Simplify metadata cache entry comparison routine (used when flushing entire cache out). - Only enable metadata cache statistics when H5AC_DEBUG is turned on, instead of always tracking them. - Simplify address comparison macro (H5F_addr_eq). - Remove redundant metadata cache entry protections during dataset creation by protecting the object header once and making all the modifications necessary for the dataset creation before unprotecting it. - Reduce # of "number of element in extent" computations performed by computing and storing the value during dataspace creation. - Simplify checking for group location's file information, when file has not been involving in file-mounting operations. - Use binary encoding for modification time, instead of ASCII. - Hoist H5HL_peek calls (to get information in a local heap) out of loops in many group routine. - Use static variable for iterators of selections, instead of dynamically allocation them each time. - Lookup & insert new entries in one step, avoiding traversing group's B-tree twice. - Fixed memory leak in H5Gget_objname_idx() routine (tangential to performance improvements, but fixed along the way). - Use free-list for reference counted strings. - Don't bother copying object names into cached group entries, since they are re-created when an object is opened. The benchmark I used to measure these results created several thousand small (2K) datasets in a file and wrote out the data for them. This is Elena's "regular.c" benchmark. These changes resulted in approximately ~4.3x speedup of the development branch when compared to the previous code in the development branch and ~1.4x speedup compared to the release branch. Additionally, these changes reduce the total memory used (code and data) by the development branch by ~800KB, bringing the development branch back into the same ballpark as the release branch. I'll send out a more detailed description of the benchmark results as a followup note. New internal API routines: Added "reference counted strings" API for tracking strings that get used by multiple owners without duplicating the strings. Added "ternary search tree" API for text->object mappings. Platforms tested: Tested h5committest {arabica (fortran), eirene (fortran, C++) modi4 (parallel, fortran)} Other platforms/configurations tested? FreeBSD 4.7 (sleipnir) serial & parallel Solaris 2.6 (baldric) serial
2003-01-10 01:20:03 +08:00
H5RS_decr(src_path_r);
H5RS_decr(dst_path_r);
H5RS_decr(canon_src_path_r);
H5RS_decr(canon_dst_path_r);
[svn-r5947] Purpose: Code cleanup Description: Clean up ID->name code: - Reformat to better match library coding standard - Changed several algorithms to be more efficient - Integrated into library more smoothly Platforms tested: eirene w/FORTRAN & C++ arabica w/FORTRAN modi4 w/FORTRAN & parallel sleipnir
2002-09-25 22:50:49 +08:00
} /* end if */
break;
default:
HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, FAIL, "invalid call");
} /* end switch */
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
done:
[svn-r6266] Purpose: Code cleanup/new feature. Description: Split FUNC_LEAVE into API and non-API specific versions. This allows a solution to compiling this branch with C++, as well as reducing the size of the binaries produced. Platforms tested: FreeBSD 4.7 (sleipnir) w/serial, parallel (including MPE) & thread-safe
2003-01-11 04:26:02 +08:00
FUNC_LEAVE_NOAPI(ret_value);
[svn-r5904] Purpose: Added 'ID to name' support Description: There is a new API function H5Iget_name Most of the changes are on H5G.c , regarding the symbol table entry struct H5G_entry_t which has 2 new fields 'name' and 'old_name' A new private function was introduced H5G_ent_copy, that does a deep copy between 2 symbol table entries The test file is getname.c Platforms tested: windows 2000, Linux, Solaris
2002-08-29 02:34:12 +08:00
}
Reference in New Issue Copy Permalink