Newsgroups: comp.parallel.mpi
From: wcs@win233.nas.nasa.gov (William C. Saphir)
Subject: Re: Something strange: f90 array subscripts
Organization: NASA Ames Research Center/MRJ
Date: 22 Aug 1996 15:58:27 -0700
Message-ID: <v3tg25ft4a4.fsf@win233.nas.nasa.gov>


The reply to this problem was right on the mark. Thought I'd add a bit
more info. There are some difficult problems in making Fortran 90
array sections work with non-blocking MPI routines. There are
currently no MPI bindings for Fortran 90, so there is no guarantee
that things like array sections and derived types will work with MPI.

The MPI Forum is currently looking at Fortran 90 issues. The likely
outcome in the short term (IMHO) is that we will not be able to solve
this problem (which is fundamental incompatibility between the way
Fortran 90 and MPI work) but at least we will list things that will
work and will probably not work. For more info, you might want to take
a look at (and comment on, to mpi-bind@mcs.anl.gov) the current MPI-2
draft which is (for the moment) at
  http://www.cs.wisc.edu/lederman/mpi2/mpi2-report.ps.Z . 
Please remember that this is a work in progress and may
change substantially. 

Bill Saphir
MRJ/NASA Ames


