FYI. The MPI Forum is re-convening to make some corrections to the
MPI-2 standard. Note that this is *not* a call for new features to
MPI; it's a call for help in fixing the bugs in the current MPI-2
spec. Bill Gropp's e-mail below briefly describes the current state
of affairs and gives a URL containing more information (including how
to join the mpi-21 mailing list).
NOTE: Pardon the posting to multiple lists simultaneously -- the
intent is to spread this message far and wide to find people who are
interested in helping with defining MPI (be warned: Rusty Lusk
correctly characterized this as "grubby, grubby work"). These four
lists share a common listserver; it should be smart enough to only
send this message once to people who are subscribed to more than one
of the lists in the "To:" line.
Begin forwarded message:
From: William Gropp <gr...@mcs.anl.gov>
Date: February 9, 2007 10:35:18 AM EST
To: mpi...@mpi-forum.org
Subject: MPI 2.1
Reply-To: mpi...@mpi-forum.org
Welcome to the MPI 2.1 discussions
The MPI 2.1 web page is available at http://www.mpi-forum.org/
mpi2_1/. This page contains a link to the current errata
discussion (please use this link through the forum page, as we may
move this page in the future). You will find on this page many
issues that have been raised and are still open, as well as a draft
ballot for the third round of errata items. All items are open for
discussion and new items may be submitted to mpi...@mpi-forum.org
or to mpi-comme...@mpi-forum.org.
The first order of business is to determine the rules for voting.
The rules that the MPI Forum has been using are posted on the MPI
2.1 web page (http://www.mpi-forum.org/mpi2_1/); the history of
errata votes (linked off of that page) shows which institutions
have participated in the last two ballots. I suggest that we
discuss updates to these rules to address the fact that we probably
won't have face-to-face meetings, at least in the beginning. Then
the current voting members (based on the last two votes) can vote
on that rule change.
Discussions of the technical issues can start at any time (in fact,
the page of issues has been available since before the original
errata discussions). Send mail about these issues to mpi-21@mpi-
forum.org.
Bill
--
Jeff Squyres
Server Virtualization Business Unit
Cisco Systems
