Description: Added -K c99 flag to sxc++ flags to ienforce C99 standard;
added cache values for getrusage and gethostname
Solution:
Platforms tested: SX-6
Misc. update:
Bug fix
Description:
Forgot to remove a debug statement before committing.
Solution:
The script printined out what it thought CC_BASENAME was for debugging.
Removed this print statement.
Platforms tested:
minor change, mcr
Misc. update:
Code cleanup.
Description:
g++ on sleipnir complained and exited failed because the
malloc statement in ALIGNMENT was not properly casted.
Solution:
added the proper (char * ) casting.
Platforms tested:
Passed heping using gcc.
Sleipnir g++ passed too but that was also because the default development
mode is used.
Description: Quincey added the condition branch (else if (*((ST*)S) != (ST)((DT)(*((ST*)S)))))
in the definition of the macro H5T_CONV_Fx_CORE. It handles a special situation
when the source is "float" and assigned the value of "INT_MAX". Compilers do roundup
making this value "INT_MAX+1". This branch is to check that situation and return
exception for some compilers, mainly GCC.
The branch if (*((ST*)S) > (DT)(D_MAX) || (S_PREC < D_PREC && *((ST*)S) ==
(DT)(D_MAX))) is for some compilers like Sun, HP, IBM, and SGI where under
the same situation the "int" doesn't overflow.
A test was added to dt_arith.c to check 2 conversions, from float to int where the
float is INT_MAX, and from double to signed char where the double is SCHAR_MAX.
Platforms tested: h5committest and fuss.
Bug fix
Description:
Same bug as was just checked in for v16. Do the complicated
derived datatype check for compilers named mpiicc in addition to
mpicc.
Solution:
Platforms tested:
Misc. update:
Bug fix
Description:
Recent group code changes tweaked the size of the free space in the file
for this test. Update to the latest correct value.
Platforms tested:
h5committested
bug fix
Description:
during the generation of some test files, H5Fclose was not called
during the #ifdef detection of the scale ofsset filter, a wrong macro symbol was used
Solution:
Platforms tested:
linux
Misc. update:
Code cleanup/reorganization
Description:
Merge back some more changes extracted from the "compact group" set.
This bunch cleans up and prepares the H5G_* routines for eventual import of
new features.
Platforms tested:
FreeBSD 4.11 (sleipnir)
Linux 2.4
Mac OS X.4
Code cleanup
Description:
Merge back changes from "compact group" work that improve the
infrastructure of the library and may impact others. In this round of
merging, that includes:
- Move datatype allocation into single internal routine, instead of
duplicated code that was spread out in a dozen or so places.
- Clean up guts of object header routines (H5O_*) to allow for some of
the fancieroperations that need to be performed on groups, along with
some general improvements.
- Added a new error code
- Some minor cleanups in other code....
Platforms tested:
FreeBSD 4.11 (sleipnir)
Linux 2.4
Mac OS X
bug fix.
Description:
In Janus (Tflops), the test would seg-fault in production (compiled
with -O). If compiled with -O0, it would be okay. Track down the
problem to the for loop index i & j being declared with type hsize_t
which is usually unsigned long int. Guess the Tflop compiler has
a bug in optimization that could not handle unsigned int as for loop
index.
Solution:
Since it is no need for i & j to be hsize_t and it makes more
sense to have them with type int. Changed them to int type.
Platforms tested:
Passed in tflops and colonelk here.
Description: Added support for "big" REAL and DOUBLE PRECISION types.
Solution:
Platforms tested: heping with g95 (-r8, -d8)
mir with ifc (-r16, -r8)
Misc. update:
Description: Added support for "big" REAL and DOUBLE PRECISION
datatypes (i.e. sizes of Fortran real and double precision
may be 8 and/or 16 bytes; usually specified by compilers
flags like -r8, -r16, -d8, etc.)
Solution: Added code to generate all necessary datatypes on a fly.
Platforms tested: heping with g95 (-r8, -d8)
mir with ifc (-r16, -r8)
Note: multi file test failed when REAL is 128 bit.
I will address this later.
Misc. update:
Description: Added code to generate sizes of Fortran REAL and
DOUBLE PRECISION types. This will "almost" eliminate
H5f90i.h file that defines C stubs datatypes.
Solution:
Platforms tested: heping with g95 (-r8, -d8 and default settings)
Misc. update:
Maintenance on Windows
Description:
Add definition for int <-> float conversion exception
Solution:
Platforms tested:
MSVS 6.0 on Windows XP
Misc. update:
upgrade.
Description:
A new feature md5 is added to the release script. Upgrade snapshot
to include md5 in the snapshot release process.
Platforms tested:
No test since the change is pretty simple and it is hard to test
without really doing a snapshot release.
Misc. update:
Bug fix
Description:
Incorrectly terminated the received string in getenv_all
Solution:
Tasks would incorrectly terminat the environment string they received from the
root task in getenv_all.
They did the equivalent of:
env[strlen(str) + 1] = '\0'.
This resulted in a single "garbage" character inserted
at the end of the string. This wasn't noticed until now because it seems that on most
platforms this garbage character was a 0 anyway.
This has been corrected to;
env[strlen(str)] = '\0'.
Platforms tested:
Red Storm (where the bug was discovered)
heping (pp)
Misc. update:
Feature - md5 checksum
Description:
Added md5 checksumming to snapshot releases. When bin/release is run,
it will by default create a .tar archive and a .md5 checksum of that
archive in the destination directory.
Solution:
md5 is now a method in bin/release, along with the various ways to compress
hdf5. The checksum is always of the .tar archive, even when other forms
of compression are used (this is so that a given release always has only
one checksum, for simplicity).
Platforms tested:
mir (change to bin/release only)
Misc. update:
Bug fix
Description:
Cygwin was failing (seemingly only when I built in parallel) when it couldn't
find or build H5detect.
Fixed a dependency.
Solution:
H5init.c was depending on H5detect instead of H5detect$(EXEEXT). On Cygwin,
applications end in .exe !
Platforms tested:
mir, cygwin (finrod)
Description: Currently we are trying to match excatly Fortran and C types.
Unfortunately, some systems (SX-6) provide compiler switches
(-ew) that make ALL Fortran types to be at least of size 8.
As a result, Fortran library cannot be compiled.
Solution: Allow INTEGER(HID_T) type to be 8 bytes if necessary. The values
of this type are originated in the C library and can be safely
passed back and forth. Please note that we cannot do the same
for INTEGER(SIZE_T) type. Fortunately there is a switch
that allows size_t be 8 bytes for both C and Fortran compilers.
Platforms tested: It is a minor change, therefore SX-6 only;
daily tests will do the rest :-)
Misc. update:
Description: Put in a new API funciton, H5Tis_hard. It checks whether
the conversion function from a native type to another native type is a
compiler (hard) conversion. Also checked a test in test/dt_arith.c.
Platforms tested: h5committest and fuss.
Bug fix
Description:
Configure should now throw an error when C++ or threadsafe is enabled
and a parallel compiler is being used.
Platforms tested:
sleipnir and modi4
Tested
--enable-cxx --enable-parallel --disable-cxx
--enable-cxx --disable-parallel
cases in particular.
Misc. update:
Bug fix.
Description:
The incompatibility of enable-parallel and enable-cxx was not correctly
coded. It would complained even for the case of
./configure --enable-cxx --disable-parallel.
Solution:
Changed it to use "$enable_XYZ" = "yes". Fixed the same for the
enable-threadsafe checking too.
Platforms tested:
tested in heping.
Misc. update:
Bug fix.
Description:
Some tests showed the filesize was not as expected. But the error was
intermittent. This was a racing condition as some processes finish
extend_chunked_dataset() sooner than others and return to the main
body which proceeds to call the next test which also uses the same
test data file and alters it. That messes up the "slower" processes
which then see unexpected filesize.
Also, the routine create_chunked_dataset() which creates test data file
actually was executed by all processes. That is wrong.
Solution:
Added a barrier at the end of extend_chunked_dataset to make sure all
processes are done with the test data file before returning.
Changed create_chunked_dataset such that only one process would create
the test data file. The rest does nothing but just wait for it to finish.
Platforms tested:
Tested in TG-NCSA in which the errors were detected.
Misc. update:
Bug fix & code cleanup
Description:
Fix problem with detecting precision overflows in a more portable way.
(I was depending on the compiler not removing some temporary variables, which
was a bad bet... :-)
Also, hoist the checking for the exception handling routine out of the
inner conversion loop, which should speed up conversions which don't have an
exception handling routine defined.
Platforms tested:
FreeBSD 4.11 (sleipnir)
h5committest
Bug fix & code cleanup
Description:
Address most of datatype conversion exception handling bug that Ed
Hartnett reported. (He's reported a different problem now, but we're closer
at least).
Also, condense exception handling #ifdef's into one location instead of
spread out in so many places.
Platforms tested:
FreeBSD 4.11 (sleipnir)
Too minor to require h5committest
Bug fix#281
Description:
Committed a wrong copy in the previous checkin.
Solution:
Checked in the right one and did some code cleanup, rearrangment.
Platforms tested:
heping pp.
Misc. update:
Bug fix
Description:
Check in fix that John found for bug in H5FD_extend() code - adjust
beginning as well as size of free block when "extending" another block of
space in the file.
Platforms tested:
FreeBsd 4.11 (sleipnir)
Too minor to require h5committest
Code cleanup
Description:
Changed some scale+offset filter symbols from H5_SO_... -> H5Z_SO_...
to indicate that they are in the H5Z package.
Platforms tested:
FreeBSD 4.11 (sleipnir)
Too minor to require h5committest
Bug fix
Description:
configure used 'uname -o' to detect Cygwin. This command is fine on
Cygwin, but causes errors on some other platforms (sol).
Fixed this.
Solution:
Used 'uname' with no -o option to identify Cygwin.
Platforms tested:
mir, cygwin (on finrod)
Feature: checking configure flags (bug #90)
Description:
Some configure flags don't work together. Configure now exits with
a sensible error message if one of these combinations is specified.
Solution:
The following flags will throw errors:
--enable-cxx and --enable-parallel
--enable-cxx and --enable-threadsafe
--enable-threadsafe and --enable-parallel
--enable-fphdf5 and --enable-fortran
Platforms tested:
mir, modi4, heping
Misc. update:
Description: Changed configure's default disable-dconv-exception and
disable-dconv-accuracy back to enable-dconv-exception and
enable-dconv-accuracy. The test has been finished.
Platforms tested: fuss - tested before.
Description: Changed the names of macros defined in H5Tpkg.h from
H5T_CONV_INTERIM_XXX_XXX to H5T_CONV_INTERNAL_XXX_XXX to be more
descriptive.
Platforms tested: fuss - simple change
Bug fix
Description:
Errors on sol may have been due to "make clean" being run both
manually and by automake. Removed the manual run.
Also cleaned c++/examples/ii_files directory created by some compilers
(e.g., on modi4).
Solution:
This does mean that examples directories will not be cleaned by
'make clean'.
Platforms tested:
sol, mir, heping, modi4
Updating the filter based on H5Pset_scaleoffset API change.
Description:
Still more changes are needed for the filter for floating-point.
Solution:
Platforms tested:
Misc. update:
Updating scale offset filter.
Description:
Using enum to represent scale_offset type.
Change the order of parameters of H5Pset_scaleoffset.
Solution:
Platforms tested:
sol 2.9, Liunx 2.4, AIX 5.1
Misc. update:
