[Mpi3-bwcompat] Stealing the working group

[Solt, David George]

Thanks for the suggestions.  I'm reworking the last proposals a bit.  My goal is to come up with something that allows for most interfaces (all but f77) to have a way to do large counts, leave f77 alone, and still have some consistency across languages and across what we already do.   

1) Function overloading as available in Fortran90 and newer right?
2) One concern I have with going to MPI_Count is that I think that MPI_Send(buf, 10, ....) should always be legal.   If we introduce a type that is up to the implementation, then that may not be legal any longer?  Do we specify in the standard that MPI_Count MUST be defined by the implementation to be something that can be cast to/from an integer constant?
3) Another concern with introducing MPI_Count TYPE in Fortran90/newer is that it is inconsistent with the way other types are handled (MPI_Comm, MPI_Request, MPI_Info, etc.)   It seems a bit strange to me that we are going to pick out this one thing and make it a TYPE in all newer Fortrans even though everything else is declared as an INTEGER.   I guess we are breaking new ground either way because we have nothing that is declared as an INTEGER*8 either, but using some form of an intrinsic type seems like less of a jump to me than making one thing a TYPE's and leaving the rest as a built in type.
4) What will MPI_Request and other pre-existing concepts be declared as in the F2003 bindings?  

Thanks again,
Dave

-----Original Message-----
From: mpi3-bwcompat-bounces_at_[hidden] [mailto:mpi3-bwcompat-bounces_at_[hidden]] On Behalf Of Jeff Squyres
Sent: Thursday, January 07, 2010 7:44 AM
To: MPI-3 backwards compatability WG
Subject: Re: [Mpi3-bwcompat] Stealing the working group

Gah -- the meeting time and information fell off my calendar.  Can you resend?  I may have already inadvertently scheduled a conflict...  :-(

-----

Notes on slides:

Slide 5: can you explain the con "...since it may double memory usage"?

Slide 7: might want to mention Fortran function overloading, a la C++ (which would alter the spirit of the "Fortran (if implementation is using 32-bit counts)" comment on slide 8) -- i.e., this could be a F2003 bindings-and-onward solution only...?

Slide 9: Might want to re-title this -- I think you're saying what I said in my above bullet (only fix the "large" count issue for the F2003-and-beyond bindings), but your title says "Drop Fortran 77 support", which taken out of content (and it will be ;-) ), will create an uproar because people will assume that you're dropping F77 support altogether.  You didn't mean that, right?  :-)

Slide 10: I don't think that never having used TYPE in Fortran before is a con -- we're going to use it in the F2003 bindings, for example.

Slide 10: I'm not sure why 3b is different than 3a (or even 2) -- why not have an MPI_COUNT type *and* use fortran/c++ function overloading?  int/INTEGER counts can be deprecated -- push users towards MPI_COUNT over time.

Slide 11: is that your actual recommendation?  If so, I'm confused by bullet #2...

(I'm probably reflecting my bias of using fortran function overloading here...)

On Jan 5, 2010, at 2:04 PM, Solt, David George wrote:

> Hi all,
> 
> Ok, MPI_Count/long send counts is the backward compatibility WG's charter, but I'm hoping that this is a good set of the Forum to run my slide deck by.  I'm trying to put together a status and a proposal of this issue for the next meeting so that we don't spend 50% of our time there talking about this again with no resolution.   (50% of our time would be ok if we came to a resolution).  I won't be there so I'm also going to look for volunteers to present this.  I may get someone from Platform Computing, since they are ex-HP-MPI people who initially created this ticket, but if someone is really interested to present it, let me know.  So, if you could look at these slides and give me feedback, that would be great!
> 
> Also, don't forget we are schedule to meet this coming Friday.
> 
> Thanks,
> Dave
> 
> <MPI_Count.pptx>_______________________________________________
> Mpi3-bwcompat mailing list
> Mpi3-bwcompat_at_[hidden]
> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-bwcompat


-- 
Jeff Squyres
jsquyres_at_[hidden]
_______________________________________________
Mpi3-bwcompat mailing list
Mpi3-bwcompat_at_[hidden]
http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-bwcompat