From: nissej@XXXXXXXXXXXXXXXXXXX
X-Mailer: exmh version 2.1.1 (devel)
Reply-To: nissej@XXXXXXXXXX
To: mpi-core@XXXXXXXXXXX
cc: Richard Frost <frost@XXXXXXXX>,William Gropp <gropp@XXXXXXXXXXX>
Subject: Re: Fwd: Re: MPI_GET_PROCESSOR_NAME \0 or not? 
Date: Wed, 15 Dec 1999 10:08:59 +0100


Although I think one should also agree on what is to happen when the 
supplied character string is longer than MPI_MAX_PROCESSOR_NAME since 
the standard says "at least of length MPI_MAX_PROCESSOR_NAME".

As a Fortran programmer the corresponding equivalent is then thought of 
as a CHARACTER*(*) in which case it is actually possible to suspect that 
the subroutine will modify more than MPI_MAX_PROCESSOR_NAME in the process
of blank filling the argument. I think this would be a preferred 
behaviour but that it maybe needs to be clarified that this is what is
supposed to happen or this will also be a possible point of different 
implementations doing different things to the argument.

/Nils