On Apr 30 2010, Ralph Castain wrote:
> Guess this has been too upsetting a question - I'll work off-list with
> the other developers to determine an appropriate OMPI behavior.
I have responded to Ralph Castain by Email, but I need to correct the
implication in the above.
What OpenMPI chooses to do, and to document, is a matter for OpenMPI alone.
The MPI standard makes it very clear that implementations are encouraged to
produce sane specifications for the behaviour that MPI has left undefined,
and to document them.
What OpenMPI proposes for inclusion in the forthcoming MPI 3.0 standard has
to be acceptable to all of the other MPI implementations, present and
future. Regrettably, Ralph Castain's desired specification is not feasible
in general, and is therefore unsuitable for MPI 3.0 (unless the portability
of MPI is sacrificed).
So there are two issues, which shouldn't be confused.