[MPIWG Fortran] MPI_SIZEOF: redux

[Jeff Hammond]

Lots of user codes are using mpif.h, regardless of their Fortran dialect. Google for it some time and then multiply by a factor to account for Fortran users' lack of use of version control.

Maybe you don't have lots of users or customers to deal with, but some of the rest of us do, and deprecating mpif.h and "use mpi" is not going to fly with them. 

Do the MPI C bindings violate K&R? They definitely work with C89, which is older than F90, so age arguments are not valid here. I'd love it if mpi had C99 and C11 bindings, but it doesn't. 

Jeff

Sent from my iPhone

On Jun 5, 2014, at 8:54 PM, "Malcolm Cohen" <malcolm at nag-j.co.jp> wrote:

>> Is this intrinsic in F90? Are we deprecating F77 and F90 bindings? Else we need this.
> 
> (a) Fortran 77 was superceded as a standard 25 years ago.  Do you have bindings to K&R (pre-ANSI) C?
> 
> (b) The MPI bindings do not even come close to adhering to the Fortran 77 standard.
> 
> (c) No, this intrinsic was new in Fortran 2008.  OTOH, it is possible to calculate "sizeof" in Fortran 90 without much effort (STORAGE_SIZE is merely a convenience function).
> 
> But I agree it is rather early to be deprecating MPI_SIZEOF just because of STORAGE_SIZE, since quite a few compilers don't have it yet.
> 
> Cheers,
> -- 
> ................................Malcolm Cohen, Nihon NAG, Tokyo. 
> _______________________________________________
> mpiwg-fortran mailing list
> mpiwg-fortran at lists.mpi-forum.org
> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpiwg-fortran