[MPIWG Fortran] [MPI Forum] #393: Fortran module CONTAINS - Advice to implementors

[Bill Long]

On May 10, 2014, at 12:18 PM, MPI Forum <mpi-forum at lists.mpi-forum.org> wrote:

> 
>   ''Advice to implementors.''  An implementation that provides a Fortran
> interface
>   must provide a combination of MPI library and module or include file
> that uses the
>   specific procedure names as described in Table 1.1 on page 5. This
> requirement
>   prohibits implementing MPI modules with the CONTAINS Fortran keyword
> such that
>   the MPI routines are not interceptable. ''(End of advice to
> implementors.)’'

I would certainly object to being told I’m not allowed to provide an alternate module with better performance   I’d be OK with it if the last sentence were changed to

“This does not prohibit implementing alternative MPI modules with the CONTAINS Fortran keyword such that the MPI routines are not interceptable when these alternative MPI modules are used.”

The current wording suggest that implementors are not allowed to provide extra (better) features beyond what is required, but instead are forced to only provide the worst performing implementation, with no alternative.  This is bad for everyone.  (Except for the guy with the slow implementation who does not want competition.  One HOPES that is not the reason here.)  I think the alternate wording I gave above would be a clear resolution of any questions about the use of CONTAINS, and is in the best interest of MPI overall.


Cheers,
Bill


Bill Long                                                                       longb at cray.com
Fortran Technical Suport  &                                  voice:  651-605-9024
Bioinformatics Software Development                     fax:  651-605-9142
Cray Inc./ Cray Plaza, Suite 210/ 380 Jackson St./ St. Paul, MN 55101