find the shared library support option in r6623.

version.

is needed for suncc.

Reviewed by gropp.

appropriate variables are declared for it to work in emulation mode as
well. This showed up as an error with the suncc compiler on linux
which is not natively supported by OPA.

error out at configure time, instead of just throwing a warning and
failing at make time.

[svn-r6613] Fixed the size of long double complex C++ type - The type size is used for special C++ types NOT MPI types to support fortran types. This fixes the failing arcomplex C++ test on windows.

fixed bug caused by insufficient check in adding new CommProcThdID in CommProcThdIDMap. The bug is caused by intercomm event in clog2 file. Intercomm event contains both IntraCommCreate and LocalIntraComm etype event which were
not distinguisable in CommProcThdIDMap.  Different CommProcThdIDs are indistinguisable in CommProcThdIDMap causes replacement of CommProcThdID's isUsed
value from true to false.

[svn-r6602] Handling unhandled defns - No ckpt available on windows & fix the name change of var to enable ckpt

[svn-r6600] Ignoring C attribute alignment issues - part of multiple aliases support - for now on windows. See r6268 for details

duplicate environment at the proxy, so the bootstrap filtering being
done at the UI didn't really take effect.

environment and overwrite it at the remote node, while allowing each
bootstrap server to specify exceptions. For example, ssh can specify
DISPLAY as an exception, and SLURM can specify all SLURM_ variables as
exceptions.

together with POLLIN instead as return values for poll().

Reviewed by buntinas@.

known failure.

command arrives in multiple socket reads.

seconds.

[svn-r6584] added ckpt_continue() and ckpt_precheck() netmod functions, removed ckpt_shutdown() netmod function, changed ckpt_restart() to return business card as parameter to be consistent with netmod_init()

These channels are old and have been deprecated for a while.  Their
presence is hampering PMI API development, so I have removed them.
These were the only channels seriously using the "process locks" code,
so that mess has also been deleted.  The only remaining useful
functionality (MPIDU_Yield) has been moved to the OS wrappers.

specify what its expected output string is when an application
terminates with a timeout. I've updated it with the current string
Hydra returns.

proxy just dumps output to stderr, there is no header prepended, so
the UI can't understand the string.