in the ompi_progress() function. Not yet fully operational, as there
is something going on with the subscription not getting called. E-mail
coming to those interested.
This commit was SVN r5343.
the configure-time param checking variable, rather than "true". Removes
the warning about not being able to enable MPI param checking during
MPI_INIT.
This commit was SVN r5341.
"-x" or "-X" already in the agent executable command line, and
debugging is not enabled, then automatically add "-x" to the agent
command line to disable X forward (and avoid locking issues with
$HOME/.Xauthority, etc.).
This commit was SVN r5331.
- Fix bug 1270 (which incorrectly identified the rsh pls as the
problem): if we "mpirun -np 2 ./test", we would find /bin/test if it
appeared in the PATH earlier than ".". Now, if there is a "/" in
the executable name anywhere, don't do a PATH search to find the
executable.
This commit was SVN r5330.
especially upon abnormal termination of a process. Not yet integrated
into the fork pls; pending more discussion with other developers.
This commit was SVN r5326.
just dont create the ptl_peer struct for this node. Later the PML have to take in account the
management of the reacheable nodes.
This commit was SVN r5314.
OMPI_ERR_INIT_FINALIZE macro (meaning that MPI_INIT has not yet been
invoked or MPI_FINALIZE has already been invoked), then the
communicator argument that is passed to ompi_mpi_abort() is
meaningless and should not be consulted.
This commit was SVN r5313.
overhead to the critical path - but ensures buffer reference counts are released
when requests are not released by the application
This commit was SVN r5311.
let's just have a short for loop that does effectively the same thing,
and gives us better guarantees about [not] modifying the source
buffer, and that one definitely has to free() the output when done
(the implementations of basename() on different platforms vary on
these guarantees). :-)
This commit was SVN r5307.
Monday with Craig:
- split two f90 modules into two separate .f90 files so that
dependencies can be satisfied properly in parallel builds (i.e.,
mpi_kinds.f90 can be compiled and its f90 module can be generated
before any others are compiled, because it is needed by all of the
other f90 source files).
- rename mpi.i.h* to be mpi-f90-interfaces.h, just to be a little more
clear and less name-confusing with mpi.h and mpif.h
- update the build process in src/mpi/f90, including printing out a
warning that compiling mpi.f90 may take quite a while :-\
- update the wrapper compilers to add in the Right Things for
compiling F90 MPI applications
- add a check in configure to find what flag the F90 compiler uses to
identify where f90 module files live
This commit was SVN r5297.
will have their own compiler's bootstrapping, which we do not want to
include in the C-only libmpi (these were placeholders anyway).
This commit was SVN r5296.
