Bug Fix
Description:
The 'carry' variable in the general floating point-to-floating point
conversion routine wasn't getting initialized correctly and was causing
problems while converting certain sequences of floating point numbers
on the Cray SV1.
Solution:
Added 'else carry=0;' line to always reset the carry variable to a known
value.
Platforms tested:
Cray SV1 (killeen)
2001-02-03 00:19:25 Robb Matzke <matzke@llnl.gov>
* H5T_conv_f_f: Fixed a bug where I had forgotten to increment the floating-point
exponent if rounding the significand resulted in a
carry. Thanks to Guillaume Colin de Verdiere for
finding this one!
Bug Fix
Description:
Generic property code was leaking memory when removing a property from a
property list class in certain situations.
Solution:
Remove property correctly from class in all situations.
Platforms tested:
Solaris 2.7 (arabica)
Bug Fix
Description:
Fix definition of strdup on Windows machines.
Solution:
Merged duplicate definitions into one cohesive block of definitions.
Platforms tested:
FreeBSD 4.2 (hawkwind)
Bug fix
Description:
Certain combinations of hyperslabs, especially those that have to be
strip-mined during I/O were causing data to be incorrectly transferred.
Solution:
Changed the code in H5S_get_hyper_regions to be more careful about the
regions of the current dimension that are valid. Sometimes, regions which
had already been iterated through were being re-processed.
Platforms tested:
FreeBSD 4.2 (hawkwind)
Bug fix (sort of)
(H5FDgass.c: just deleted an old code. Nothing to do with this fix.)
Description:
When --enable-debug=all was used, couple routine printf warnings
to stderr that it is an expensive operation. This messed up
the testing of h5ls since it included stderr output in its expected
output and did not expect the extra messages.
Solution:
Changed it such that the extra messages are printed only
if the appropriate h5-debug-class setting is on (via environment
variables, for example).
Platforms tested:
eirene (debug=all), modi4(parallel,-64,default configure).
Suppress warnings for release.
Description:
Always declare strdup as extern, figure out way to avoid warnings later.
Platforms tested:
Linux 2.2.16 (eirene)
Bug fix
Description:
Datasets were allowed to be created with chunks larger than the maximum
dimension for each dimension.
Solution:
Compare chunk sizes against maximum dimensions and reject dataset creations
which have chunks too large for the dimensions in the dataspace.
Platforms tested:
FreeBSD 4.2 (hawkwind)
Purpose:
Arrange codes for avoiding windows warnings
Description:
Solution:
put windows including header files into H5private.h
Platforms tested:
win 2000, confirmed at eirene
Purpose:
Avoid windows warnings
Description:
Solution:
put some windows specified headers into this file
Platforms tested:
Win 2000, confirmed by eirene
Code cleanup for C++
Description:
'xor' is a keyword in C++, causing problems compiling the library with a
C++ compiler.
Solution:
Change variables and structure fields from 'xor' to 'xor_val'
Platforms tested:
FreeBSD 4.2 (hawkwind)
Typo fix and comment
Description:
The FUNC_ENTER of H5FD_mpio_flush had a wrong function name. Corrected
it. Also filled in the missing original author name.
Added a 'this is collective.' to all collective functions.
Platforms tested:
It was just editing. Can't go wrong. :-)
Code cleanup.
Description:
Fixed _lots_ (I mean _tons_) of warnings spit out by the gcc with the
extra warnings. Including a few show-stoppers for compression on IRIX
machines.
Solution:
Changed lots of variables' types to more sensible and consistent types,
more range-checking, more variable typecasts, etc.
Platforms tested:
FreeBSD 4.2 (hawkwind), IRIX64-64 (modi4)
Bug fix
Description:
H5P_DATASET_XFER was the name from the beginning through 1.2. It was
changed to H5P_DATA_XFER on v1.3.0. Then it was changed back to
H5P_DATASET_XFER right before the release of v1.4.0-beta2.
Define an alias here to help applications that had ported to v1.3.
Should be removed in later version.
#define H5P_DATA_XFER H5P_DATASET_XFER
Also removed the v1_2 compat condition since it is a v1.3 compat
issue now.
Platforms tested:
modi4 -64 parallel.
Purpose:
several warning and bug fix for windows
Description:
different system calls on windows
Solution:
for win32, use only those system calls
Platforms tested:
windows 2000
Purpose:
a bug fix
Description:
delete the line for unused defined variables on windows.
Solution:
delete the line.
Platforms tested:
windows 2000
[machines you have tested the changed version. This is absolute
important. Test it out on at least two or three different platforms
such as Big-endian-32bit (SUN/IRIX), little-endian-32(LINUX) and
64-bit (IRIX64/UNICOS/DEC-ALPHA) would be good.]
Purpose:
a bug fix
Description:
Get rid of warning information for comments on windows(**/* */)
Solution:
add a space between ** and /*. Now it is ** /*
Platforms tested:
windows 2000
Improved debugging output
Description:
Added a little more descriptive information to the output for h5debug
when debugging B-trees in the file.
Platforms tested:
FreeBSD 4.2 (hawkwind)
Bug fixes
Description:
Fix two bugs:
- Datasets with vlen datatype which were created but not written to
were not being read back in correctly from the file.
- If an existing space conversion path was found for a conversion, it
was possible that the optimized read/write routines would be used
inappropriately.
Solution:
Patched vlen datatype conversion code to correctly handle zero-length
sequences.
Added a check to the space conversion code to make certain that the
optimized conversion routines are still appropriate when an existing
path is found.
Platforms tested:
FreeBSD 4.2 (hawkwind)
2000-12-29 08:59:03 Robb Matzke <matzke@llnl.gov>
* H5FD_stream_query: The `flags' argument is advertised to be output only. Therefore I
added code to zero its value before the feature bits
are assigned.
2000-12-29 08:57:58 Robb Matzke <matzke@llnl.gov>
* H5FD_stdio_query: The `flags' argument is advertised to be output only. Therefore I
added code to zero its value before the feature bits
are assigned.
2000-12-29 08:56:44 Robb Matzke <matzke@llnl.gov>
* H5FD_srb_query: The `flags' argument is advertised to be output only. Therefore I
added code to zero its value before the feature bits
are assigned.
2000-12-29 08:56:22 Robb Matzke <matzke@llnl.gov>
* H5FD_sec2_query: The `flags' argument is advertised to be output only. Therefore I
added code to zero its value before the feature bits
are assigned.
2000-12-29 08:55:42 Robb Matzke <matzke@llnl.gov>
* H5FD_multi_query: The `flags' argument is advertised to be output only. Therefore I
added code to zero its value before the feature bits
are assigned.
2000-12-29 13:26:28 Robb Matzke <matzke@llnl.gov>
* H5FD_mpio_write: Avoid MPI_Get_count() when using the LAM implementation of MPI.
This code must have other problems anyway because Kim
Yates has a similar kludge in place to avoid negative
return values from the same function.
2000-12-29 13:25:30 Robb Matzke <matzke@llnl.gov>
* H5FD_mpio_read: Avoid MPI_Get_count() when using the LAM implementation of MPI.
This code must have other problems anyway because Kim
Yates has a similar kludge in place to avoid negative
return values from the same function.
2000-12-29 11:28:13 Robb Matzke <matzke@llnl.gov>
* H5FD_mpio_flush: Read and rewrite the last allocated byte of the file to insure
that EOF>=EOA.
2000-12-29 09:32:49 Robb Matzke <matzke@llnl.gov>
* H5FD_mpio_write: Removed unused variable.
2000-12-29 09:32:15 Robb Matzke <matzke@llnl.gov>
* H5FD_mpio_write: Fixed a failure return value that should have been FAIL instead
of NULL.
2000-12-29 09:31:47 Robb Matzke <matzke@llnl.gov>
* H5FD_mpio_read: Removed unused variable.
2000-12-28 16:09:31 Robb Matzke <matzke@llnl.gov>
* H5FD_mpio_query: The `flags' argument is advertised to be output only. Therefore I
added code to zero its value before the feature bits
are assigned.
2000-12-29 08:55:06 Robb Matzke <matzke@llnl.gov>
* H5FD_log_query: The `flags' argument is advertised to be output only. Therefore I
added code to zero its value before the feature bits
are assigned.
2000-12-29 08:54:15 Robb Matzke <matzke@llnl.gov>
* H5FD_gass_query: The `flags' argument is advertised to be output only. Therefore I
added code to zero its value before the feature bits
are assigned.
2000-12-29 08:53:43 Robb Matzke <matzke@llnl.gov>
* H5FD_family_query: The `flags' argument is advertised to be output only. Therefore I
added code to zero its value before the feature bits
are assigned.
2000-12-29 08:52:45 Robb Matzke <matzke@llnl.gov>
* H5FD_dpss_query: The `flags' argument is advertised to be output only. Therefore I
added code to zero its value before the feature bits
are assigned.
Updating debugging information for new features, etc.
Description:
Lots of the new features added to the library during the last year or so
have been added without updating the debugging routines used by h5debug.
Solution:
Added more of the new features (although not all of them) to the debugging
routines for h5debug. Also included some more information to expand on
the information printed from h5debug.
Platforms tested:
FreeBSD 4.2 (hawkwind)
Bug Fix
Description:
The #ifdef ... #else statement had no line breaks in it. This
caused the compiler on Sangamon to freak out saying the #ifdef
was invalid.
Solution:
Put line breaks in it.
Platforms tested:
Sangamon
Removal of spurious warnings
Description:
On HP-UX systems, the compiler would warn of a redeclaration of
H5G_entry_t in the H5Sprivate.h header.
Solution:
Removed the prototype declaration since H5Oprivate.h includes the
H5Gprivate.h header file already.
Purpose:
a windows bug fix
Description:
so that sec2 driver can support >2 G file on windows NTFS
Solution:
using lseek_i64 for windows platform
Platforms tested:
tested on windows 2000, confirmed on eirene and arabica
