[svn-r20872] Issue 7564 - Allow H5Tcreat to create string type (fixed-length and variable-length). I also added a test case in dtypes.c.
Tested on jam and linew - relatively simple.
[svn-r20870] Description: Fixed HDFFV-7605: Dim Scales APIs- error when unattaching dim. sacles
* Fixed Loop counter when removing object in dimension set with more then one scale.
Tested: jam (intel and gnu)
[svn-r20869] Purpose:
HDFFV-7514 - GMQS: h5dump - improve effectiveness of testing scripts for handling expected output files
Description:
Fist phase for h5dump test update. Next phase will be for h5ls.
This is prerequisite for 'HDFFV-7602 Tools - upgrade tools' test scripts to share test files among tools, providing framework for test file re-usability.'
Tested:
jam (linux32-LE), koala (linux64-LE), heiwa (linuxppc64-BE), linew (solaris-BE), Windows (32-LE), ...
[svn-r20862] Adjust CPack configuration location by removing version extension in folder name. Add install configuration commands if ext libs are packaged with project.
[svn-r20855] Issue 2599 - Cross-test BE and LE data. I added the test for deflate, szip, fletcher32, shuffle and nbit filter in cross_read.c and gen_cross.c. I also took out the data file for OpenVMS because the data isn't native but IEEE for floating-numbers.
Tested on koala, heiwa, and jam.
[svn-r20804] Purpose:
HDFFV-7594 - GMQS: h5dump shows wrong values for H5T_STD_I8LE dataset on Blue-gene (ppc64 linux BE +@)
Description:
For the STD_I8 type data, updated to use 'signed char' instead of 'char' as
'char' could be defined as unsigned on a certain system, which cause h5dump
to display data values incorrectly.
Tested:
jam (linux32-LE), koala (linux64-LE), heiwa (linuxppc64-BE), tejeda (mac32-LE)
[svn-r20796] Refactor print_data_region_blocks/ponts to remove context passed in by value. The print functions needed an independent context with only the current indent level passed in by argument. Also synched the two routines logical flows and corrected a mem_space error.
h5dump (and h5stat, h5jam/h5unjam) help option not tested. Added h5dump parse_command_line technique for h5stat and h5jam/unjam and created testfiles.
Tested: local linux
[svn-r21103] Modified test code to work around a pgcc optimizer bug encountered
on Koala.
Tested on: Koala -- PGI -- debug and production
Jam -- gcc, intel, PGI -- debug and production -- with and without AIO
Phoenix -- gcc -- debug and production, with and without AIO.
[svn-r20797] Resolved failures on bsd machines by modifying configure.in to not add
-D_POSIX_C_SOURCE=199309L to CC flags for BSD.
Updated to trunk revision r20795.
Tested phoenix (serial -- aio and aio with error recovery)
jam (serial -- gcc, intel, pgi, aio force on and off)
freedom & loyalty (serial)
[svn-r20778] I changed the macro definition of DIR_SEPC and DIR_SEPS to ']' for OpenVMS in H5private.h. Strictly speaking, it isn't
the directory seperator. But it can simplify the code and make the H5L_build_name function cleaner.
Tested on jam because the change only affects VMS.
[svn-r20776] Refactor print_data_region_blocks/ponts to remove context passed in by value. The print functions needed an independent context with only the current indent level passed in by argument. Also synched the two routines logical flows and corrected a mem_space error.
Tested: local linux
[svn-r20771] Problem:
Test scripts sometimes need to filter some system-specific messages from the
actual output so that it can match the correct expected output. These filtering
functions, ususally called "STDOUT_FILTER()" and "STDERR_FILTER()" were being
repeated in individual test scripts. This becomes a maintenance problem and
is error prone.
Solution:
Extract the two filter functions code to bin/output_filter.sh and then each
test script sources it in. This allows reuse of coding and i...
[svn-r20767] Purpose:
HDFFV-5928 - GMQS: h5diff problem and improvement on comparsing the same objects
Description:
Improved performance by eliminating duplicated action for getting object
information in half from the previous fixe when comparing group vs group.
This is addition to the previous commit r20676.
Tested:
jam (linux32-LE), koala (linux64-LE), heiwa (linuxppc64-BE), tejeda (mac32-LE), linew (solaris-BE), cmake
[svn-r20759] I commented out 2 test cases using H5_CANNOT_OPEN_TWICE because they open the same file twice. OpenVMS
can't do it. Other systems still run them.
Tested on jam. Simple change.
[svn-r20756] Fixes h5diff test failure on BE systems.
H5Dread() produces different output for invalid enum type values
on systems of different endiannes. On BE systems, it always
emits -1. On LE systems, the invalid value is emitted. This
difference caused the h5diff test to fail since it tests to
see if different invalid enum values are reported as different
(on BE systems they are not since -1 == -1).
This fix removes the "invalid vs. different invalid" test
while we fix the library bu...
[svn-r20724] Purpose: Fix a bug in h5diff when enum values are compared that do
not represent a valid enum value.
Description:
The h5diff code compares enum values by converting them to strings and
then comparing them. When the enum value is out of range and can't be
converted to an enum string representation, the comparison was skipped.
The code now flags differences when one of the two enum values is out
of range (two out of range values are compared in memory). A test has
been...
[svn-r20720] Merge changes to scripts to run installed examples from hdf5 1.8.7 r 20718. The
default prefix in the scripts will use the compile scripts in the installed bin
directory.
[svn-r20713] Packed Bits Merge:
removed option defines and #ifdef/#endif
refactored all printf to HDfprintf in h5dump.c
formatted and indention improvements
synched with 1.8 branch
Tested: local linux
[svn-r20716] Updated from trunk and resolved conflicts.
tested on phoenix (serial -- with and without aio)
jam (serial -- gcc, intel, and PGI compiler, with and without aio)
[svn-r20691] Purpose: Fix problem exposed by external link cache
Description:
When closing an object that had two open id's, the library would previously fail
to account for the status of the "holding_file" field, potentially causing the
file's "nopen_objs" field to become inconsistent. This caused problems when
opening the same object twice through an external link using th external file
cache. Modified the library to properly account for the status of the
"holding_file" field when closin...
[svn-r20689] Changed compiler name from xlf/mpxlf_r to xlf90/mpxlf90_r to avoid
misunderstanding. xlf/mpxlf_r are actually Fortran 77 compilers. HDF5 requires
Fortran 90 or higher.
[svn-r20686] Added "xlc_r" as a recognized XL compilers.
Tested:
NCSA BP AIX system. No need of h5committest because the changes affect AIX
systems only.
[svn-r20676] Purpose:
- HDFFV-5928 - GMQS: h5diff problem and improvement on comparsing the same objects
Description:
Fixed:
1) adding h5tools_is_obj_same() function to check if two given IDs or paths point to the same object. This function can be very useful for other tools and applications.
2) using h5tools_is_obj_same() at h5diff() and diff() in h5diff.c. If two paths point to the same object, there is no need to check the details of the object since we know there is no di...
[svn-r20666] Bug fix: VMS has trouble to deal with absolute path. I changed dtypes.c to use the file name only. I also
updated the current directory path in vms/test/H5srcdir_str.h.
Tested on jam and OpenVMS.
