On Feb 18, 2016, at 1:31 PM, Jeff Hammond <[email protected]> wrote:
Please tell me why we want to do something different for Fortran.
Seems to be a good point. What is mpich doing if there is no accompanying Fortran compiler?
Full mpi.h from my Clang MPICH builds with FC=false attached. Relevant excerpt below. /* Fortran types */
#define MPI_COMPLEX ((MPI_Datatype)MPI_DATATYPE_NULL)
It looks like OMPI does a mix of setting some Fortran datatypes to (effectively) MPI_DATATYPE_NULL and others to what we guess/assume they should be equivalent to (e.g., MPI_INTEGER --> MPI_INT).
Middleware that compiles but cannot be used in any nontrivial way because MPI_DATATYPE_NULL may (at least with Open-MPI) cause the program to abort. I don't see a problem in breaking compilation when successful compilation today leads to unsuccessful execution.
+1 -- Jeff Squyres [email protected] For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/