This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "MPICH primary repository". The annotated tag, v3.0.3 has been created at b6ef7ec081ed86814fed393fee1f03bae8fadce8 (tag) tagging 21750800ed2475c43e291dd78c0a285ea97babec (commit) replaces v3.0.2 tagged by Dave Goodell on Thu Mar 28 15:59:10 2013 -0500 - Log ----------------------------------------------------------------- tagging 'v3.0.3' Antonio J. Pena (1): Fixed MPI_Get_elements returning wrong values Dave Goodell (23): clmake: support more patterns tt#1754: fix warnings in ROMIO external32 code explain external32 limitations in RELEASE_NOTES clmake: more patterns and a regex fix fix Fortran MPI_WEIGHTS_EMPTY initialization hydra: squash loop warning from modern clang ROMIO: fix buffer/datatype mismatch in Isend/Irecv clmake: use `\W` instead of `\b` clmake: filter "install-sh" too .gitignore: fixup for [006dd97e] move `_NULL` handle definitions earlier in mpi.h fix compile error when `long double` is disabled configure: add `--disable-long-double` `long double _Complex` depends on `long double` clang annotations: fix `long double _Complex' mpi.h: guard `long double`-derived MPI types Merge branch 'fix-long-double' tests: gate `long double _Complex` fix `long double _Complex` logic from [a9305613] fix `complex<long double>` bug from [3f1fcead] `--enable-g=all` now means _all_, add `=most` update CHANGES for v3.0.3 release bump version numbers for v3.0.3 release James Dinan (48): Fix type incompat. for size_t in BSend degug msgs BUGFIX: Unlock piggybacking for Get_accumulate Eliminate enqueueing of lock op in RMA ops list PERF: Flush should ignore inactive targets PERF: Overlap req/wait for lock-all on shr men win Fix incorrect timing reporting in llist bench Minor cleanups to MCS mutex test Added missing thread safety to Comm_compare Fixed several warnings in lockcontention3 test Cleaned up test compiler warnings Temporarily reverted is_gacc_op bugfix Consolidated RMA op finalization code Converted packet type field to 16-bits Added flags field to RMA op packet headers Integrated flags into RMA op processing Added flags to MPID_Request RMA sync. piggybacking from origin->target Simplify get completion Cleanup of FOP packet header Removed unused single_op_opt field from MPID_Request Flush piggybacking Implemented lock op piggybacking for MODE_NOCHECK Add a MODE_NOCHECK version of the llist bench test Cleaned up error names for CH3 RMA implementation Finished adding memory fences for shr mem win Remove unnecessary flush for empty epochs Added missing CPPFLAGS in rma tests makefile Merge branch 'packet-header-flags' Added missing descriptions for new RMA err classes Added missing function boilerplate decls Moved comm_split_type override from CH3 to Nemesis CH3 default shared memory window implementation Moved window flavor error checking to public API Merge branch 'shr-mem-win-ch3' Added missing CS ENTER/EXIT guards to type_commit Updated changes file with RMA changes for 3.0.3 Updated req-based ops to handle PROC_NULL Cleaned up handling of PROC_NULL in RMA ops Added new operations to RMA PROC_NULL test Fix handling of 0-byte xfers in req-gen ops Test handling of 0-byte RMA xfers Cleanup comment in rmazero test Fix req. op. completion outside of PT epoch Add additional wait/test cases to Req. ops. test Fix Irecv race in wait/test_all/any/some err tests Replace calls to sleep with MTestSleep in tests Fix coll6 test hanging with more than 10 processes Updated CHANGES file Pavan Balaji (6): Bug-fix: Allow configure to check for cobalt as well. Added support for Cobalt to the CHANGES file. Upgrade to hwloc-1.6.1rc2. This version uses libpciaccess instead of Merge hydra-specific hwloc patches from Updated the CHANGES file with the hwloc-1.6.2rc1 information. Bug-fix: correct MPIR_proctable setup when we use wraparound ranks. Ralf Gunter (1): README: fix pamid configuration example Rob Latham (1): feature-encode NTFS quirk William Gropp (5): Added support for passing the length of a Fortran CHARACTER as size_t instead of int. Added a configure option to set; note that despite documentation to the contrary, it looks like some compilers still pass the length as an int (the common practice for decades), so changing to size_t must be made with extreme caution Added needed tests based on coverage analysis Added tests for user-defined error handlers in Fortran Files that were modified to supprt the tests of user-defined error handlers from Fortran More files for user error handler tests ----------------------------------------------------------------------- hooks/post-receive -- MPICH primary repository