Newsgroups: comp.parallel.mpi
From: llewins@msmail4.hac.com (Lloyd J Lewins)
Subject: Re: non-blocking error code
Organization: Hughes Aerospace Electronics Co.
Date: Thu, 22 Jun 1995 08:59:32 -0800
Message-ID: <llewins-2206950859320001@x-147-16-95-58.es.hac.com>

> Therefore, if anyone out there knows with authority what
> error class a "out of system resources" report from MPI_Isend would be, it
> would be a great help if you would share the info.
> 
> I think it would be worth documenting in the MPI standard, since in my
> experience, this kind of problem occurs fairly often with extensive use
> of non-blocking communication.

The standard is clear - there are no guarantees that you can continue after
the error you are discussing - 2.7 page 13 line 21

   "This document does not specify the state of a computation after an
    erroneous MPI call has occurred"

and 7.2 page 191 line 33

    "After an error is detected, the state of MPI is undefined. That is, using
     a user-defined error handler, or MPI_ERRORS_RETURN, does not neccessarlily
     allow the user to continue to use MPI after an error is detected."

Thus, if you write a program which by design generates errors, you cannot expect
it to be portable. I suggest that you implement your own higher level flow
control to ensure that you don't consume all the system resources.

--------------------------------------------------------------------------
Lloyd J Lewins                                  Mail Stop: RE/R1/B507
Hughes Aerospace and Electronics Co.            P.O. Box 92426
                                                Los Angeles, CA 90009-2426
Email: llewins@msmail4.hac.com                  USA
Tel: 1 (310) 334-1145
Any opinions are not neccessarily mine, let alone my employers!!

