Buffers that will be written to disk will now always be cleared since
not doing this has huge security implications.
Tested on: 64-bit Ubuntu 15.10 (Linux 4.2.0, x86_64) gcc 5.2.1
serial autotools
parallel autotools (MPICH 3.1.4)
serial CMake
- remove H5AC_ind_dxpl_id and use only H5AC_dxpl_id everywhere instead.
- remove flush_me_collectively flag from cache entries
- add a collective sanity check (MPI_Barrier) for every HDF5 API routine that could possibly touch the file. This is trigerred when the environment variable H5_COLL_API_SANITY_CHECK is set to a non 0 digit.
tested on BB-8 with serial and parallel.
I missed changes to a couple of expected error output files with the last
checkin.
Tested on:
MacOSX/64 10.11.2 (amazon) w/serial & parallel
(h5committest forthcoming)
Normalize against incoming changes from the phdf5_metadata_opt branch.
Tested on:
MacOSX/64 10.11.2 (amazon) w/serial & parallel
(h5committest forthcoming)
Description:
User Adam Rosenberger reported a failure when using the member function
AbstractDs::getArrayType(). This problem was caused by missing
initialization of the ArrayType's members in some cases.
Solution:
- Added ArrayType::setArrayInfo() to retrieve rank and dimensions of
an array and store them in memory for easy access.
- Re-factored a few functions to use the new function.
- We'll give him 1.8.16 patch
Platforms tested:
Linux/32 2.6 (jam)
Linux/64 (platypus)
Darwin (osx1010test)
The current implementation for Mac OSX platform is a kludge that sometimes left behind mounted volumes.
Temporary screen out the checking of installtion_check step (so that daily test may pass).
Tested: by hand running cmakehdf5 in platypus and osx1010test.
Switch filter plugin to use H5allocate_memory and H5free_memory instead of
malloc and free.
Tested on:
Linux/32 2.6.x (jam) w/serial
Linux/64 2.6.x (jam) w/serial
MacOSX/64 10.11.2 (amazon) w/serial & parallel
Switch v2 B-tree min & max records to be void*'s to be cleaner and generate
less warnings.
Tested on:
Linux/32 2.6.x (jam) w/serial
MacOSX/64 10.11.2 (amazon) w/serial & parallel
Add --enable-memory-alloc-sanity-check option to configure, to track and
sanity check memory allocations within the library. This is orthogonal to the
--enable-using-memchecker option and can be used with/without it.
Tested on:
MacOSX/64 10.11.2 (amazon) w/serial & parallel
(h5committest forthcoming)
Clear memory for shared message info when "deferring" and using memory
checker, so valgrind doesn't whine.
Tested on:
MacOSX/64 10.11.2 (amazon) w/serial & parallel
(h5committest forthcoming)
Bump the nunmber of v2 B-tree records for the random 4-level B-tree update
test by another 25,000 records.
Tested on:
MacOSX/64 10.11.2 (amazon) w/serial & parallel
(Too minor to require h5committest)
Clean up many mismatches between malloc/free and H5MM_malloc/H5MM_xfree
in the library and tests (and use of H5free_memory and H5Dvlen_reclaim). Also
make H5Ocopy use a private version of H5Lexists, which doesn't internally
throw (and suppress) errors when an object (or the path to it) isn't found in
the destination.
Tested on:
MacOSX/64 10.11.2 (amazon) w/serial & parallel
(h5committest forthcoming)
Minor code neatening and cleanups (same as r28755) on revise_chunks branch.
Tested on:
MacOSX/64 10.11.2 (amazon) w/serial & parallel
(Too minor for h5committest)
Bump number of records for level 4 B-tree test, so that it's more likely to
always get to level 4.
Tested on:
MacOSX/64 10.11.2 (amazon) w/serial & parallel
(h5committest forthcoming)
Add 'update' operation to v2 B-trees, which will insert a new record, or
modify an existing one, depending on if the record exists or not.
Tested on:
MacOSX/64 10.11.2 (amazon) w/serial & parallel
(h5committest forthcoming)
Bring over rest of performance improvements for extending chunked datasets
and normalize against revise_chunks branch.
Tested on:
MacOSX/64 10.11.2 (amazon) w/serial & parallel)
(h5committest forthcoming)