Configuration feature
Description:
'make install' now tests both static and shared libraries if both are installed.
Solution:
Previously, shared libraries were only tested when static libraries were not installed.
Also cleaned up line in commence.am that was including HL library in all Makefiles.
Platforms tested:
mir (Makefile change only)
testing feature
Description:
HDF5 Test Express change; now accepts values 0 to 3.
Solution:
As explained in the RFC and documented in the code, there are now four
levels of "express-ness", 0 (no limit) to 3 (1 min limit).
It is now up to test writers to take advantage of the various levels
to tune their tests' run times.
Platforms tested:
mir (minor change)
Description: Added a command file for testing tools; it has a framework
to do exactly (almost) what UNIX tools test scripts do:
- run test tool and save output
- compare output with expected result
- figure out the difference and
- print PASSED /*FAILED* in a manner similar to the UNIX output
It needs to be populated with more tests and reused for other tools.
Solution:
Platforms tested: VMS server
Misc. update:
Description: Data conversion from long double to (unsigned) long long
returns some incorrect values on Mac OS 10.4 and SGI IRIX64 6.5. The
conversions start to go wrong when the long double is
20041683600089727.779961 (0x4351ccf385ebc8a0bfcc2a...). If adjusting
the values higher by assigning 0x...c8a0cf... or 0x...c8a0df..., the
converted values go wild.
Solution: Detect this error in configure.in and disable compiler
conversions from long double to (unsigned) long long for Mac OS 10.4
and IRIX64 6.5.
Platforms tested: h5committest, Mac OS 10.4, and IRIX64 6.5.
Code checkpoint
Description:
Add in more new features for the fractal heap code, mostly bringing in
more ability for deleting objects (which isn't completely working yet).
Also, checkpoint free space manager code, which is essentially complete
(although it needs some more work after the metadata cache has some additional
features)
Platforms tested:
FreeBSD 4.11 (sleipnir)
Linux 2.4 (chicago)
h5committest
New feature
Description:
Add '%t' format for HDfprintf() to print htri_t (and hbool_t) values.
Platforms tested:
FreeBSD 4.11 (sleipnir)
Linux 2.4 (chicago)
h5committest
Bug fix
Description:
Don't attempt to merge object header messages when a file is opened
read-only.
Mark the object header dirty when messages are merged, as well as the
particular message affected.
Platforms tested:
FreeBSD 4.11 (sleipnir)
Too minor to require h5committest
Description: Added new tests for the h5stat tool; added --version flag
to display library version information and updated expected
output files.
Solution:
Platforms tested: heping, shanti
Misc. update:
Description: Added new tests for the h5stat tool; added --version flag
to the h5stat tool to display library version information
Solution:
Platforms tested: heping, shanti
Misc. update:
Description: Added test script and two test files in the tools/misc/testfiles
directory for testing h5stat tool; modified Makefile.am in tools/misc
and configure.in files to run the new test script; reran bin/recofigure
to update relevant configuration files; updated MANIFEST
Solution:
Platforms tested: heping with in place and srcdir builds; will do more testing
after the checkin.
Misc. update:
Bug fix
Description:
Fix bug which could lead to files with incorrect count of messages in the
header for an object. The exact sequence of operations is complicated and
is described in the release notes.
Solution:
Mark merged null header messages as dirty.
Platforms tested:
FreeBSD 4.11 (sleipnir)
Linux 2.4 (chicago)
Mac OS/X (amazon)
Bug fix
Description:
Changed the semantics of --enable-stream-vfd and --enable-shared.
Both are now disabled by default when parallel HDF5 is being built.
Solution:
Both shared and stream-vfd are disabled by default in parallel, but can
be enabled explicitly if the user chooses.
Before, shared was always enabled by default and stream-vfd was
always disabled in parallel.
Platforms tested:
heping, mir (configuration change only)
Bug fix - bugzilla #552
Description:
On Cray X1, trying to use : as an argument confused the system.
Solution:
Added a test in configure to see if : as an argument is bad.
If so, skipped the test.
Platforms tested:
mir, Cray X1 (change to configure only)
Update HDF5 Fortran90 type definition file on windows to
reflect the changes in Linux.
Description:
INT has been changed to Fortran_INTEGER in the file.
Solution:
Make the change.
Platforms tested:
Visual fortran 6.6c with MSVS 6.0
Only affects windows.
Misc. update:
Bug fix.
Description:
It failed when only 1 is used to test.
Solution:
Cleanup the code a little so that it works for any number
of processes to invoke it.
Platforms tested:
h5committested,
Description:
Fixed typos and re-arranged some functions to be in sync with 1.6.
Platforms tested:
Linux 2.4 (heping)
SunOS 5.9 (shanti)
AIX 5.1 (copper)
Description: I am checking in changes for the h5stat tool:
- updated Makefile.am to use tools library
- regenerated Makefile.in
- added command flags to display selected information
- added support for external files and filters
- small output formatting changes
For more information about current features and output see
updated h5stat RFC http://hdf.ncsa.uiuc.edu/RFC/h5stat/h5stat-spec.pdf
Solution:
Platforms tested: mir, heping, shanti, copper 64-bit, copper 32-bit parallel
using files from tools/testfiles directory and some NASA files
from our Web site.
Misc. update:
Some collective chunk IO macro names are confusing, change them to more meaningful
names.
Description:
H5Pset_dxpl_mpio_chunk_opt will set a flag so that the library can do one linked IO or mul
ti-chunk IO with collective in chunking storage directly. That is, the library won't do an
alyses to determine this.
The flags for the enum type we used before are:
H5FD_MPIO_OPT_ONE_IO
H5FD_MPIO_OPT_MULTI_IO
They are not good names because of the following two reasons:
1. It doesn't reflect chunking storage
2. OPT is kind of redundant and misleading,
Solution:
We change the names to
H5FD_MPIO_CHUNK_ONE_IO
H5FD_MPIO_CHUNK_MULTI_IO
Platforms tested:
Since only macro names are changed, no need to test with h5committest.
Misc. update:
Some collective chunk IO macro names are confusing, change them to more meaningful
names.
Description:
H5Pset_dxpl_mpio_chunk_opt will set a flag so that the library can do one linked IO or multi-chunk IO with collective in chunking storage directly. That is, the library won't do analyses to determine this.
The flags for the enum type we used before are:
H5FD_MPIO_OPT_ONE_IO
H5FD_MPIO_OPT_MULTI_IO
They are not good names because of the following two reasons:
1. It doesn't reflect chunking storage
2. OPT is kind of redundant and misleading,
Solution:
We change the names to
H5FD_MPIO_CHUNK_ONE_IO
H5FD_MPIO_CHUNK_MULTI_IO
Platforms tested:
Since only macro names are changed, no need to test with h5committest.
heping(mpich 1.2.6)
Misc. update:
Bug fix.
Description:
Sometimes the parallel prefix is given in the form of nfs:/mnt/pfs which
if given to a non-MPIO VFD like the default H5Fcreate, it would fail.
Called h5_rmprefix which would return the non-prefix component
part of the file name which would be okay for the default H5Fcreate and
such.
Platforms tested:
Tested in heping parallel.
Feature
Description:
Sometimes the parallel prefix is given in the form of nfs:/mnt/pfs which
if given to a non-MPIO VFD like the default H5Fcreate, it would fail.
Added a new routine h5_rmprefix which would return the non-prefix component
part of the file name which would be okay for the default H5Fcreate and
such.
Platforms tested:
Tested in heping parallel.
Description: Ran /bin/h5vers to change suffix to alpha4 after creating hdf5-1.8.0-alpha3
tar ball and tagging the release with hdf5_1_8_0_alpha3 tag
Solution:
Platforms tested:
Misc. update:
Misc. update: Document some errors in test/dt_arith.c when converting data from
long double to (unsigned) long long and from (unsigned) long long to long double.
Description: Updated VMS installation instructions to reflect support
for the deflate filter; documented scale+offset test failures
when G_FLOAT is used (not IEEE double)
Solution:
Platforms tested:
Misc. update:
Description: There was some logical confusion about the value of the macro
H5_HW_FP_TO_LLONG_NOT_WORKS. It's supposed to be true for Windows, but false
for Unix machines.
Solution: Corrected.
Platforms tested: fuss(Linux) and amazon(MacOS 10.4) - simple change.
Description: dsets and objcopy tests were missing from the command files
Solution: Added appropriate commands to compile and run tests
Platforms tested: VMS server with native float
Misc. update:
Description: Windows doesn't recognize "long long" but "long_long".
Solution: The library defined a more general type "long_long". Use
it instead.
Platforms tested: fuss - simple change.
Added Exress Test mode.
Description:
The test takes a long time to run. It tempted developers to skip the
h5committest.
Solution:
Added Express Test mode to allow tester an option to run it in express
mode (shorter time.) There are already encoded QAK macro which skips
many tests. Follow that pattern to skip tests when HDF5TestExpress is
set.
Platforms tested:
Tested in heping, both serial and parallel.
feature.
Description:
Added the option of -64 which is similar to -n32. This is needed by
Atlantia which has the ABI as -n32.
Platforms tested:
Pretty straight forward but not testable by h5committest.
Will be tested when daily tests run.
Feature
(Code and tests are done by Christian. I just committed the code.)
Description:
Rewrote the purpose of this test. Now it tests these three cases,
/*
* Test following possible scenarios,
* Case 1:
* Sequential create a file and dataset with H5D_ALLOC_TIME_EARLY and large
* size, no write, close, reopen in parallel, read to verify all return
* the fill value.
* Case 2:
* Sequential create a file and dataset with H5D_ALLOC_TIME_EARLY but small
* size, no write, close, reopen in parallel, extend to large size, then close,
* then reopen in parallel and read to verify all return the fill value.
* Case 3:
* Sequential create a file and dataset with H5D_ALLOC_TIME_EARLY and large
* size, write just a small part of the dataset (second to the last), close,
* then reopen in parallel, read to verify all return the fill value except
* those small portion that has been written. Without closing it, writes
* all parts of the dataset in a interleave pattern, close it, and reopen
* it, read to verify all data are as written.
*/
Platforms tested:
Tested in copper, tg-ncsa and heping, all in parallel mode.
Bug fix
Description:
tg-login was having an error about H5match_types.Tpo being missing.
Solution:
tg-login was trying to build H5_match_types twice at the same time and
becoming confused as a result.
Massaged the dependencies in the Makefile.am to prevent this.
Platforms tested:
mir, sol, tg-login
1) Check in potential fix to unreproduceable bug in t_cache observed on
Cobalt
2) Check in code supporting added pinned entry features in the metadata
cache.
Description:
1) Elena encountered a bug in t_cache when running on Cobalt. On
investigation I found a potential race condition in t_cache that
could explain her observation.
2) Quincey requested additions to the metadata cache allowing renaming
of pinned entryies, and marking of pinned or protected entries as
dirty.
Solution:
1) Modified t_cache.c to prevent the potential race condition. Elena
was unable to reproduce the bug on Cobalt, so we don't know if my
fix actually addressed the problem.
2) Added code supporting the requested functions. The changes were
relatively trivial, but required substantial new test code.
Platforms tested:
h5committest, serial test on heping, parallel test on phoenix.
Misc. update:
Description: dtype_file.txt is a text file for testing complicated compound type
for H5LTtext_to_dtype(). dtype_file_readable.txt is just a human-readable version
of dtype_file.txt for debugging purpose.
Platforms tested: No test.
Description: The current code in Yacc parser doesn't handle big-endian
machine for enum type correctly.
Solution: Depending on the super type of enum, use appropriate integer
type to pass in enum value.
Platforms tested: h5committest and fuss.
Description:
Shanti compiler destroy unnamed objects later than others, which caused
some reference counting test fail. Revised the test so that destructors
are called at the same time, regardless the differences of compiler
implementation.
Revised some constructors, close, operator=, and destructors to make
sure that all the object ids are handled properly.
Platforms tested:
Linux 2.4 (heping)
SunOS 5.9 (shanti)
HPUX 11.00 (kelgia)
AIX 5.1 (copper)
Description:
Shanti compiler destroy unnamed objects later than others, which caused
some reference counting test fail. Revised the test so that destructors
are called at the same time, regardless the differences of compiler
implementation.
Revised some constructors, close, operator=, and destructors to make
sure that all the object ids are handled properly.
Platforms tested:
Linux 2.4 (heping)
SunOS 5.9 (shanti)
HPUX 11.00 (kelgia)
AIX 5.1 (copper)
fix.
Description:
More machines stop supporting "rsh machine command".
Worse yet, it hangs the "rsh machine command" process.
Solution:
Switch the order of testing remote execution to use
ssh first, then rsh.
Platforms tested:
no h5committest since it does not test this command.
Change is trivial and will monitor Daily test tonight.
Code checkpoint
Description:
Add a few minor tweaks & corrections to the fractal heap code.
Platforms tested:
Linux 2.4 (chicago)
Too minor to require h5committest
