New Feature
Description:
Added new H5Dfill() routine to fill the elements in a selection for a
memory buffer with a fill value. This is a user API wrapper around some
internal routines which were needed for the fill-value modifications
from Raymond as well as Pedro's code for reducing the size of a chunked
dataset.
Platforms tested:
FreeBSD 4.5 (sleipnir) [and IRIX64 6.5 (modi4) in parallel, in a few
minutes]
Purpose:
New feature
Description:
Added a query function H5Tget_member_index for compound and enumeration
data types, to retrieve member's index by its name.
Platforms tested:
Linux 2.2
Minor h5ls bugfix for compound types with array members.
(Similar changes shortly for 1.4 branch).
2002-04-02 10:22:55 Robb Matzke <matzke@arborea.spizella.com>
* H5F_istore_read: Realigned a couple of comments.
Bug Fix & Code Cleanup
Description:
The MPI-IO optimized transfer routines
(H5S_mpio_spaces_read/H5S_mpio_space_write) are not being invoked in all
the cases where they could be used.
Additionally, the code for determining if an optimized transfer is wrapped
into the actual I/O transfer routine in a very confusing way.
Solution:
Re-enabled MPI-IO optimized transfer routines in all the cases where they
should work.
Extracted all the pre-conditions for optimized transfers into separate
routines from the transfer routines.
Platforms tested:
FreeBSD 4.5 (sleipnir) & IRIX64 6.5 (modi4)
Features.
Description:
Error stack used to report only hdf5 predefined error messages
because it takes only static strings. Runtime defined messages
were not pushed to the stack.
Added the means and macros to push MPI error strings onto the
hdf5 error stack. Added a new minor error class as H5E_MPIERR
for this class of messages.
H5Epulbic.h, H5E.c:
Added H5E_MPIERR and its minor class description.
H5Eprivate.h:
Added HMPI_XXX macros to push MPI error strings to the stack.
H5FDmpio.c:
Changed couple places to use the new macros to test the new
macros. A more through changes to make use of these new
macros will be done later.
Platforms tested:
eirene (serial, parallel)
modi4(parallel)
Autotools Update
Description:
I've updated autoconf, automake, and libtool to the latest/greatest
versions; 2.53, 1.6, and 1.4.2 resp.
Many changes come with the new versions:
- ltconfig is no longer used
- acconfig.h is no longer used (#define values are declared with
the macro)
- regeneration of all of the aclocal.m4, configure, and
H5config.h.in files.
- new config.{guess,sub} files
- new ltmain.sh file
Platforms tested:
AIX (blue), and Linux
Code cleanup
Description:
The function pointer for doing parallel I/O was being set in the H5D_read
and H5D_write routines instead of in H5S_conv, where it belongs.
Solution:
Added an extra 'flags' parameter to H5S_find to allow the function to
determine when to set the MPIO functions instead of the regular optimized
I/O routines and get rid of code messing with the function pointer in the
H5D routines.
Platforms tested:
IRIX64 6.5 (modi4)
Bug Fix & Feature
Description:
The selection offset was being ignored for optimized hyperslab selection
I/O operations.
Additionally, I've found that the restrictions on optimized selection
I/O operations were too strict and found a way to allow more hyperslabs
to use the optimized I/O routines.
Solution:
Incorporate the selection offset into the selection location when performing
optimized I/O operations.
Allow optimized I/O on any single hyperslab selection and also allow
hyperslab operations on chunked datasets.
Platforms tested:
FreeBSD 4.5 (sleipnir)
Bug Fix (sorta)
Description:
"small" contiguous hyperslabs were not being detected correctly by
H5S_hyper_select_contigous() leading to poorer performance than possible.
Solution:
Corrected check for small hyperslabs. Also cleaned up the H5S_find() code.
Platforms tested:
FreeBSD 4.5 (sleipnir)
Purpose:
Bug fix(#697)
Description:
Variable-length string wasn't treated as string.
Solution:
Added character set and padding into VL string type.
Platforms tested:
FreeBSD
Merge of name component length limit bugfix from 1.4 branch.
A couple of very minor typos fixed.
2002-03-18 14:30:58 Robb Matzke <matzke@arborea.spizella.com>
* H5B_assert: Fixed typos in comments.
Purpose:
fix a windows bug
Description:
1. To create a >4GB file, fstati64 has to be used instead of fstat.
change fstat to fstati64 for WIN32 macroes. Some discussions are going on
the general issue on how to better handle with WIN32 or other similar platform.
This check-in is just a reminder not to forget the windows bug.
2. erase a WIN32 macro(include winsock2.h) at h5detect.c. It is not a problem now.
Solution:
use fstati64 instead of fstat for win32.
Platforms tested:
windows2000(confirmed at linux machine eirene)
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: ----------------------------------------------------------------------
Bug Fix
Description:
When reading a contiguous hyperslab that spanned the entire dataset and
was larger that the type conversion buffer, the hyperslab routines need
to fill the type conversion buffer and then return to the I/O routines.
When the I/O routines resume the hyperslab operation, it was possible to
have a combination of coordinates which caused the hyperslab iterator
to incorrectly advance in the file, causing some data to be re-read or
re-written.
Solution:
Corrected the H5S_hyper_iter_next routine to correctly handle contiguous
hyperslabs that span the entire dataset dimensions.
Platforms tested:
Linux (eirene)
Code Motion
Description:
Removal of HDF4 from the configure/Makefiles. This is a precursor to
the actual physical removal of the HDF4 tools from the HDF5 tree.
Platforms tested:
Arabica, Dangermouse
Purpose:
added a new function H5D_update_chunck that will update the chunk information; for the moment this function
just has debugging of the b-tree
Platforms tested:
w2000 , eirene
Purpose:
code clean up
Description:
just some clean up of the functions H5D_set_extend and H5S_set_extend before the big change
(changing chunk info in the b-tree struct of the dataset)
Platforms tested:
w2000, eirene
Purpose:
Bugfix
Description:
Internally created socket wasn't closed
Solution:
In my last commit I only fixed the case for sending data over a socket.
There was also a resource leak for the receiving side where an internally
created socket wasn't closed. Same wrong logical condition, same fix.
Platforms tested:
Linux, SGI
Purpose:
Bugfix
Description:
If the socket, used for data streaming by the Stream driver, was created
by the driver itself, it did not close it after the file had been closed.
A resource leak occured.
Solution:
This bug was just introduced due to a logical condition.
Fixing the condition solved the problem.
Platforms tested:
Linux, SGI
Adding support for dll
Description:
Added the definition of __DLLCPP__ depending on:
HDF5_CPPDLL_EXPORTS: C++ API is to be exported - this name is
generated by MSVC++ when the project was created.
HDF5CPP_USEDLL: C++ API dll is to be used (imported.) Any
applications, that use the C++ API dll, must define this
name in the project setting.
On non-windows platforms, __DLLCPP__ is nil.
Platforms tested:
Linux 6.2 (eirene)
Windows 2000
Bug Fix
Description:
If a non-zero fill-value is used for a chunked dataset, any non-existent
chunked read with an "all" selection (or a contiguous hyperslab selection)
will return zero for those instead of the user's fill-value.
Solution:
Fixed I/O code to pass down fill-value to "optimized" I/O routines, so it
will be available to fill the user's buffer with.
Platforms tested:
FreeBSD 4.5 (sleipnir)
Purpose:
fixed a typo of the name H5S_set_extend on the call
FUNC_ENTER (H5S_set_extend, FAIL);
Platforms tested:
w2000 octopus
linux eirene
sun arabica
IRIX64 modi4
Bug fix
Description:
When a block was preempted from the chunk cache, it is possible that one
of the pointers in the algorithm is invalidated and would generate a core
dump.
Solution:
Re-calculate the internal pointer and move the the preemption after the
re-calc.
Platforms tested:
FreeBSD 4.5 (sleipnir)
Description:
modified the below files to inlude a new public function H5Dset_extend, similar
to H5Dextend, but it can lower the dimension
this function requires 2 more new private functions:
H5D_set_extend
H5S_set_extend
Platforms tested:
Bug fix & feature add
Description:
Added new API function H5Sget_select_type to determine type of selection in
a dataspace. Return values are defined by the H5S_sel_type enumerated type
in H5Spublic.h
Also, hyperslab operations involving a "all" or "none" selection are not
generating the correct resulting selections.
Solution:
Added more code to make hyperslab operations against an "all" or "none"
selection generate the correct results.
Platforms tested:
FreeBSD 4.5 (sleipnir)
Bug fix
Description:
Generic property classes derived from an existing class were not having
the parent class's properties copied into the new class.
Additionally, derived classes were not being detected correctly.
Solution:
Copied properties from parent class into derived class.
Modified H5P_isa_class_real() to walk back up the chain of parent classes
to proper detect derived classes.
Platforms tested:
FreeBSD 4.5 (sleipnir)
Code speedup
Description:
Chunking I/O routines are reading in an entire chunk when performing I/O
on the chunk, even if the chunk will be too large to cache.
Solution:
If the chunk is too large to cache, uncompressed and has been allocated
space in the file, or if we are using the MPI-I/O VFD, perform the I/O
directly to the chunk, instead of reading the chunk into memory, updating
it and immediately writing the entire chunk back out.
Platforms tested:
FreeBSD 4.5 (sleipnir) (using serial access) and IRIX64 6.5 (modi4) (using
parallel access with MPI-I/O)
Bug fixes
Description:
This checkin addresses two separate bugs:
#1 - When a compound datatype with a VL datatype as a field is used
to read or write data, the datatype conversion paths added to
the table of paths were treating the VL datatypes somewhat too
genericly - they weren't dicriminating between VL datatypes used
in different files and therefore there was the possibility that
a path in the table having information for one file could get used
with a second file, causing errors or core dumps (if the first
file was closed before the later file used the path).
#2 - When composite (compound, array or VL) datatype paths in the datatype
conversion table are being tested to see if they can be used for
the current datatypes being converted, they can cause additional
conversion paths to be registered in the conversion path table.
Since this causes the global table to change size and entries to
move around, it is possible that the local variables tracking a
potential path could become incorrect as the global table was
changed out from underneath them.
Both bugs fixed are described in bug #703
Solution:
Two separate fixes:
#1 - Changed H5T_cmp to differentiate between VL datatypes in different
files and not to return datatypes in two different files as equal.
#2 - Note size of global table before evaluating datatype paths. Then,
after the appropriate path has been chosen, check if the size of
the global table has changed and recompute the position of the path
chosen in the table if necessary.
Platforms tested:
FreeBSD 4.5 (sleipnir)
More optimizing for byte order conversion. Mostly just making code
easier to follow by simplifying Duff's device coding of the loops.
I also split the conversion function into two functions with different
names so output from H5T debugging indicates whether the optimized or
unoptimized case was invoked.
2002-01-25 10:48:34 Robb Matzke <matzke@arborea.spizella.com>
*: Added prototype for H5T_conv_order_opt().
More optimizing for byte order conversion. Mostly just making code
easier to follow by simplifying Duff's device coding of the loops.
I also split the conversion function into two functions with different
names so output from H5T debugging indicates whether the optimized or
unoptimized case was invoked.
2002-01-25 10:48:54 Robb Matzke <matzke@arborea.spizella.com>
* H5T_conv_order: Removed Duff's device consisting of >500 lines of code. Unrolled
two loops by hand resulting in <300 lines of code
which is a few percent faster and far easier to read.
More optimizing for byte order conversion. Mostly just making code
easier to follow by simplifying Duff's device coding of the loops.
I also split the conversion function into two functions with different
names so output from H5T debugging indicates whether the optimized or
unoptimized case was invoked.
2002-01-25 10:47:13 Robb Matzke <matzke@arborea.spizella.com>
* H5T_init_interface: Registered conversion function H5T_conv_order_opt() under two
names. H5T debugging will report the conversion
function as either "ibo(opt)" or "fbo(opt)".
Bug Fix
Description:
When file space was returned to the file space free-list for reuse,
occasionally raw data allocations which used space from the free-list
would overlap with the metadata accumulator and get over-written with
the cached information in the accumulator, corrupting the data.
Solution:
Check if the space about to be recycled on the free-list is going to be
used for raw data and also overlaps with the metadata accumulator cache,
avoiding using space that fits those criteria.
This fixes bug #701
Platforms tested:
FreeBSD 4.5 (sleipnir)
Feature improvement
Description:
Re-write how the free-list headers were used, to reduce the amount of space
added to each malloc request. Reduced header for array and block free
list items from 24 bytes to 8 bytes and eliminated the header for fixed-size
free list items entirely. This should reduce the amount of memory that the
library uses.
Platforms tested:
FreeBSD 4.5 (sleipnir) & IRIX64 6.5 (modi4)
