GitLab

causing ch3:sock errors to also be off by one, resulting in very strange
errors.

Nemesis.  In MPI-3, ch3 will require shared memory in order to support
MPI_Win_allocate_shared().

since we don't need to maintain backward compatibility here.
Hardcoded them to const, so they are better tested.

Reviewed by goodell.

[svn-r10054] Fixed bug where we were calling the barrier override function for a sub-communicator rather than the communicator itself.  Reviewed by goodell@

yet implemented below the ADI level.

[svn-r10049] Fixed bug in lmt_send_RTS where we released a request after destroying it.  This bug highlights a larger issue with how we're handling requests after an error:  In some places we're destroying them and in others we're releasing them.  fixes #1638

as a patch within MPICH2, which is no longer needed.

Reviewed by goodell.

No reviewer.

No reviewer.

user passes -help.

No reviewer.

No reviewer.

process.  The proxies do not need to distinguish between the local and
global binding options.  They just pick whatever mpiexec gives them.

No reviewer.

binding and mapping.

No reviewer.

to expose to MPICH2.

No reviewer.

[svn-r10031] Added several MPICH2_CONST qualifiers previously missed (found when compiling the MPI-3 examples), as well as adding reduce_scatter_block to MPE2 collchk

[svn-r10029] Generalizaed the remote launching of conftest without assuming the existence of a common filesystem.

[svn-r10027] bugfix for PAC_FC_AVAIL_INTEGER_MODELS during cross-compilation so it passes correct info to PAC_FC_INTEGER_MODEL_MAP.

[svn-r10026] Replaced passing of cross-variable in PAC_FC_*MODEL*() by cross-value to be consistent with how other Fortran macros pass cross-value.

automake >= 1.12.1 gives deprecation warnings about the "AM_" form,
which causes autogen.sh to bomb out in the hydra package because we have
"-Werror" set.

No reviewer.

[svn-r10021] initial update of F90 datatype tests to remove the use of PAC_RUN_LOG (in preparation for cross-compile support).

Thanks to Micha Jahn for the bug report and suggested fix.

No reviewer.

Thanks to "csadlo" for the bug report and suggested fix.

No reviewer.