h5diff issues "unable to read attribute" error message when comparing two files
Description:
h5diff did not handle attributes of variable length strings correctly. It created the wrong memory type if one but not both compared attributes are of the variable strings type. It ended in H5Aread() failure.
Solution:
We decided to treat the case when one type is variable length string and the other is not as incomparable. That bypassed the above error. We still need to figure out the proper way, if possible, to compare variable string with non-variable string types.
Tested: h5committested plus tested in Swallow (OS X 10.8).
and examples.
Platforms tested: Manual testing in place and using srcdir on jam, platypus, and emu with default and
PGI, Intel and new GNU compilers. ifort compiler was also tested with -i8 and -r8 flags
on jam. CMake tested on jam.
Remove all traces of MPI-POSIX VFD and GPFS detection/code.
Remove remaining traces of stream VFD.
Remove testpar/t_posix_compliant test (it's not actually verifying anything).
Clean up H5D__mpio_opt_possible() further.
Moved environment variable that disables MPI collective operations into
MPI-IO VFD (instead of it being in src/H5S.c).
A few other small code cleanups.
Tested on:
Mac OSX/64 10.9.2 (amazon) w/parallel & serial
KitWare recommendations: Change CMake commands to lower case.
Bump up CMake minimum versions.
Split test file into separate file for h5repack test.
Tested: local linux
Check in Mohamad's changes to support collective I/O on point selections,
along with some other minor cleanups.
Tested on:
Mac OSX/64 10.9.2 (amazon) w/parallel & serial
(h5committest forthcoming)
view objects. The addition of view objects in the fastforward project
is expected to be brough into the trunk sometimes in the future, which
is why we need to make this change.
Tested Manually on Jam and Ostrich.
Tested with h5commitest - Koala with intel compilers failed, but nothing had to do with those changes.
error on Koala: error while loading shared libraries: libirng.so
This is needed to avoid a conflict with the Read Context object to be
introduced in another Project.
No impact on users, since this is a private module, and low for
developers as the RC module use is very sparse in the library.
Tested with h5committest.
Some files are not distributed when release tarball is created by bin/release
according to MANIFEST. But --enable-build-all compiles them in test/.
Solution:
Distributed those missing files for now. Need to resolve if these files
should be distributed or not. And if not, fix --enable-build-all NOT to
compile them.
Tested: use bin/release to generate the tarball and run --enable-build-all with
it. Error is gone and all passed.
1. The file states that:
NOTE:
We are no longer actively supporting MinGW as of 1.8.5.
2. It is not tested for v1.8.11.
3. If we ever supports it again, it can be "restored" from previous
release (e.g. v1.8.10).)
