New feature.
Description:
Added routines to report on the amount of storage for:
1) 1.6 btree and heap storage info for groups
2) 1.8 btree, fractal heap storage info for groups, attributes and SOHM table
3) btree storage for chunked datasets
4) 1.8 superblock extension size.
Platform tested:
h5committested.
New feature.
Description:
Added routines to report on the amount of storage for:
1) 1.6 btree and heap storage info for groups
2) 1.8 btree, fractal heap storage info for groups, attributes and SOHM table
3) btree storage for chunked datasets
4) 1.8 superblock extension size.
Platform tested:
h5committested.
Description:
This cleans up a lot of different things we were doing in the Windows batch scripts. For one, we now use "exit /b" to better emulate functions in batch scripts. Also changed error-checking to look for negative return values as well.
This also adds initial support for Visual Studio 2005 command-line builds in hdf5build.BAT
Tested:
WinXP
Add --enable-deprecated-symbols configure option, to allow users to
remove deprecated public API symbols at configure time.
Add bin/make_vers script to bin/reconfigure actions.
Run bin/reconfigure to regenerate autotool components.
Tested on:
Mac OS X/32 10.4.10 (amazon)
FreeBSD/32 6.2 (duty)
Linux/32 2.6 (kagiso)
Initial checkin of API version macro generation script (bin/make_vers),
along with configuration file (src/H5vers.txt) and generated header
(src/H5version.h). Right now, no version macros are defined, since the main
source code and tests need to be adjusted to pay attention to the macros
defined.
Also, fixed makefile for generating error header files when used with
GNU make.
Tested on:
Mac OS X/32 10.4.10 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
Bug fix.
Description:
It used to clean out all *.h5 when done but this could cause a racing
condition error if parallel make (e.g., gmake -j ...) is used because
the "rm ... *.h5 ..." would remove *.h5 generaged and are still used
by other test programs (e.g. xx_write/xx_read).
Solution:
Removed "*.h5" from the cleanup list since the test program has already
taken care of removing temporary *.h5 generated.
Tested platform:
smirom. Not triple platforms tested since this is just a simple shell
script change.
Bug fix.
Description:
It used to clean out all *.h5 when done but this could cause a racing
condition error if parallel make (e.g., gmake -j ...) is used because
the "rm ... *.h5 ..." would remove *.h5 generaged and are still used
by other test programs (e.g. h5_write/h5_read).
Solution:
Removed "*.h5" from the cleanup list since the test program has already
taken care of removing temporary *.h5 generated.
Tested platform:
Kagiso, both serial and parallel. Not triple platforms tested since
this is just a simple shell script change.
Remove some unnecessary code that checks for fill value buffer being
NULL.
Tested on:
Mac OS X/32 10.4.10 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
Description:
Previously, there was a very intricate process for converting VS6 project files to VS.NET format for testing on the command-line. Now that we maintain VS.NET project files, this system is no longer neccessary. The same builds are now carried out in hdf5build.BAT, so the previous files are obsolete.
Removing them, and updating the MANIFEST
Tested:
WinXP
Description:
Our Windows test script was using "ERRORLEVEL" to check the diff output for our tooltests. However, FC does not actually set ERRORLEVEL with valid input. As a work around, we pipe the output into FIND, which sets the ERRORLEVEL if it doesn't find a valid string. We probably never saw this bug before because h5dump sets the ERRORLEVEL.
Tested:
WinXP.
Description:
In the new Windows autotests, hdf5build.BAT was resetting a variable used by hdf5bt.BAT. Changed so hdf5build.BAT looks to see if hdf5bt.BAT is running before it removes this variable.
Tested:
WinXP
Description:
Compilers don't destroy temporary objects at the same time, some do
immediately after use, some delay until the end of the scope. The
delay makes it difficult to test the reference count functionality.
Solution:
Enclosed the operations on objects and check reference count after
getting out of the enclosed block. This only works for the test!
I'm assuming that checking reference count is just for the library
to test that functionality; the user will not need to do that. :-/
Platforms tested:
SunOS 5.10 (linew)
Linux 2.6 (kagiso)
Description:
I'm in the process of re-writing the Windows build scripts for Visual Studio .NET (and eventually VS2005). These will be used with an autotest script, which is in the works.
Tested:
VS.NET on WinXP, although probably still needs some work.
Description:
Explain why Fortran isn't supported on MinGW and give users an "unsupported" alternative. Also re-worked the main documentation to clarify using the HDF5 DLL libraries.
Tested:
None needed.
Make h5dump more gracious and informative about objects it doesn't
understand.
Tested on:
Mac OS X/32 10.4.10 (amazon)
No need for h5committest - very minor
Modify semantics for constant messages slightly, to allow them to be
deleted, but not modified. This works around the issue where "old" style
groups are being created with the symbol table message marked constant in
the 1.6 (and prior) branch, but the 1.8 branch needs to remove the symbol table
message when the group has been changed to the "new" style (upon insertion
of a link that the "old" style can't handle).
Tested on:
Mac OS X/32 10.4.10 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
Fix problem with VL-datatype fill values when shrinking dataset's dimensions
Tested on:
Mac OS X/32 10.4.10 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
Allow H5Dset_extent() to reduce the size of dimensions that aren't unlimited
Tested on:
Mac OS X/32 10.4.10 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
with Intel 10.0 compiler.
Solution: Use -O1 optimization flag with Intel 10.0
Platforms tested: abe and kagiso with Intel 10.0 and Intel 9.1 compilers; no committest
is needed since this is a compiler specific change.
"Genericize" the datatypes for fill value testing when extending a dataset.
Tested on:
Mac OS X/32 10.4.10 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
A few more cleanups for testing [atomic] fill values and extending datasets
Tested on:
Mac OS X/32 10.4.10 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
Extract main test for extending a dataset into separate routine, to
facilitate testing different parameters.
Tested on:
Mac OS X/32 10.4.10 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
Another minor code cleanup to chunked storage "init by extent" routine.
Tested on:
Mac OS X/32 10.4.10 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
Rename Mac OS X compiler configuration file to more generic name, which
should allow Intel Macs to configure better.
Tested on:
Mac OS X/32 10.4.10 (amazon)
Finish cleaning up "initialize by extent" routine (used when shrinking a
dataset's dimensions) before addressing VL-datatype fill value bug.
Tested on:
Mac OS X/32 10.4.10 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
Description:
This makes major updates to the Windows installation documentation, to account for the switch from VS6 to .NET. Installation instructions for 6.0 can also be found in INSTALL_Windows_Short_MSVS6.0.TXT, although it's noted that we are phasing out our VS6 support.
Also added install instrctions for MinGW, to reflect our initial support. Updated MANIFEST for new file checkin.
Tested:
Instructions for each configuration have been tested.
2) Removing configuration of c++/src/H5cxx_pubconf.h file.
Reasoning: the additional pubconf file caused compilation complications, this is cleaner.
Tested: kagiso, smirom
Clean up the chunked storage "initialize by extent" routine, in preparation
for fixing another VL-datatype fill value bug.
Tested on:
Mac OS X/32 10.4.10 (amazon)
Add small interface to "wrap" a static buffer (usually on the stack), but
still allow for buffers larger than the static buffer to be allocated. This
can eliminate _many_ short-lived buffer allocations in situations where the
buffer is a predictable size (or at least a "very likely" size).
Also, some minor code cleanups, particularly in the SOHM caching code.
Tested on:
Mac OS X/32 10.4.10 (amazon)
Description: added new configuration to generate a pubconf file
in the c++/src directory, H5cxx_pubconf.h. When C++
compiler recognizes 'offsetof', the macro
H5_CXX_HAVE_OFFSETOF is defined in the new
pubconf file.
tested: kagiso, smirom
Description:
Clean up the MANIFEST a little bit. Also removed some batch files and empty folders that were added mistakenly in the last check-in.
Tested:
Too minor.
Description:
Many files added removed in last check-in. I still need to clean up a little bit, but I just wanted to get this in before the Daily Tests start.
Tested:
chkmanifest on kagiso
Desciption:
This is the first major check-in in the process of depreciating Visual Studio 6.0. We create a new directory, windows_vs6, which now holds all of our previous VS6 project files. We expect to maintain this directory for only a short time (perhaps one release), so users have time to make the change.
The ./windows directory now contains project files for Visual Studio .Net, as well as existing source and test scripts. This will allow us to greatly simplify the build process for the majority of our users.
We have created a new solution (.Net equivolent of workspace) that houses the Fortran project files. This allowed us to remove these files from the main solution, as well as streamline the conversion in the Fortran solution. It is also much easier now to convert to Visual Studio 2005 files.
Notes:
I didn't update the MANIFEST with this check-in, I will make those changes shortly.
Also, I haven't yet updated the documentation to reflect this check-in. That is going to require more work.
We will be giving a seminar in the coming weeks about .Net and how to migrate over.
Tested:
VS6/VNET/VS2005 on WinXP 32/64-bit
Refactor fill value buffer code into one location, for better long-term
maintenance.
Tested on:
Mac OS X/32 10.4.10 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
Description:
Windows does not accept the buffer size parameter of setvbuf to be less than 2. Previously, we had our fix split between two files, H5private.h and the Windows H5pubconf.h. This was a problem for MinGW, because it generates its own H5private.h. This moves the fix out of H5pubconf.h and into H5private.h.
This shouldn't affect other platforms.
Tested:
MinGW on WinXP
Clean up SOHM code, to copy message in H5HF_op call instead of separate
calls to H5HF_get_obj_len() and H5HF_read().
Tested on:
Mac OS X/32 10.4.9 (amazon)
Linux/32 2.6 (chicago)
