Description:
The test added failed in some machines because the data file contains infinity values that different machines print them differently as "inf", "INF", "Inf", ...
Solution:
Added a "ignorecase" option to TOOLTEST() to do caseless matching between generated output vs expected output. This solved most machines problem for now.
Tested: h5committest, emu by hand for both development and production modes.
But cmake built h5dump failed to read the data file. Using the same source to build h5dump by autotools produced a h5dump that can read the test data file. Don't know why cmake could not produce a correct binary.
The tool claimed it could handle 24bit images but there was no code to handle it.
(or might be there were but was removed by previous revisions.)
Also discovered that it does not accept multiple images nor -p for palette
as its user document and online help message indicated.
Solution:
Added code to verify dimension sizes are within 8 bit raster images limit and
added tests to verify the tools correctness.
Need to update user document tool.
Tested: h5committested.
Bring r26651 from autotools_rework branch to trunk:
Remove the VSNPRINTF_WORKS macro, it's working around bugs in old SGI
& HP compilers.
Tested on:
Linux/32 2.6.18 (jam) w/serial & parallel
(Daily tested on branch for 2+ days)
- Adds 'F' suffixes for most float constants.
- A few constants MUST be of type double. These now receive the long
double L suffix and are then cast to double. I do this via a new
H5_DOUBLE() macro which was added to H5private.h.
Fixes: HDFFV-9148
Tested on: h5committest
branch.
Removed the configure option that allows selective disabling of individual
internal filters (fletcher32, shuffle, etc.).
This feature mucked up the code with a lot of #ifdefs, saved very little space,
and was not scalable to a general scheme for library size reduction. We've
decided to remove the feature while we investigate a more general scheme for
decreasing the library size.
Part of: HDFFV-9086
Tested on: h5committest
Cleans up time functions in the autotools input files. Removed deprecated
gettimeofday() time zone code and configure processing. Also removed
some unused time function configure checks and defines.
Fixes: HDFFV-9083 and HDFFV-9085
Tested on: 32-bit linux, serial and parallel (jam)
OS X (kite)
Solaris (emu)
- Remove Infering parallel compilers (C and Fortran) from configure.ac
- Remove restriction to build shared with parallel
- Cleanup parallel sections in configure.ac
- remove large file support checks
- MPE fixes.
tested with h5committest.
Last changes had caused problems with Windows platform:
1. H5private.h in io_timer.h is not accessible in the distributed shared library of tools.
Need to move it to io_timer.c.
2. debug_start_stop_time() would not be available outside of h5perf. Some favor of compiler
does not like it. Screened it out for now.
3. mere "extern" is not good enough for Windows shared library. Must mark them as DLL. Need to mark them as
"H5TOOLS_DLL".
4. STANDALONE mode is broken. Screened out the code for now.
Tested; h5committest.
Move timing measurement routines used by h5perf and h5perf_serial to tools/lib
so that they can be used by other tools too.
tools/lib/io_timer.c
tools/lib/io_timer.h
tools/perform/io_timer.c
tools/perform/io_timer.h
renamed perform/io_timer.[ch] to lib/io_timer.[ch]
tools/lib/Makefile.am
tools/lib/Makefile.in
tools/lib/CMakeLists.txt
tools/perform/Makefile.in
tools/perform/Makefile.am
tools/perform/CMakeLists.txt
MANIFEST
updated due to the move.
Tested; h5committest and jam (serial)
tools/perform/sio_timer.h:
tools/perform/pio_timer.c:
tools/perform/io_timer.h:
tools/perform/io_timer.c:
Renamed pio_timer.[ch] as io_timer.[ch];
merged content of sio_timer.[ch] into io_timer.[ch] so that both h5perf and h5perf_serial
call the combined io_timer.[ch].
Moved the debug printing code segment in set_time() in [sp]io_timer.c to as debug_start_stop_time() in [sp]io_perf.c where it belongs.
Screened out set_timer_type() and get_timer_type() which were not used by any code.
tools/perform/Makefile.am:
tools/perform/Makefile.in:
tools/perform/CMakeLists.txt
Adjusted to use io_timer.c for both h5perf and h5perf_serial.
tools/perform/pio_perf.c:
tools/perform/sio_perf.c:
tools/perform/pio_perf.h:
tools/perform/sio_perf.h:
Renamed [sp]io_time_destory() to same io_time_destroy();
Moved the debug printing code segment in set_time() in [sp]io_timer.c to as debug_start_stop_time() in [sp]io_perf.c where it belongs.
tools/perform/pio_engine.c:
tools/perform/sio_engine.c:
Removed #include io_timer.h since it was already #include by [sp]io_perf.h.
MANIFEST:
Updated the merge of [sp]io_timer.[ch] into io_timer.[ch].
Tested: h5committest and jam in extra.
Benchpar.c is retired/deleted. If it is ever needed, it can be retrieved from the SVN repository
or even previous version of HDF5 source code.
Tested: jam parallel make passed.
Previously, ph5diff may hang in a test in Linux 32 platform(HDFFV-587).
No solution was available and a patch to bypass the test is implemented.
Since the company is changing domain name soon, the patch would not be valid any more.
Tests were done (repeat 100 mpiexec -np 6 ./ph5diff -v h5diff_hyper1.h5 h5diff_hyper2.h5)
and the previous hanging problem did not occur. It is likely the newer versions of
mpich no longer have this problem. Therefore, it is decide to remove this patch
to make the test script cleaner, if nothing else.
Tested: ran build and test in Jam (linux 32) and the repeat 100 command mentioned above.
All passed.
Discovered when I temporarily #defined HDfree() to a more complicated
function while investigating something.
Tested on a local linux VM. This is a very minor change.
