[Mpi-forum] Proposal for the MPI-3 standardization process

Underwood, Keith D keith.d.underwood at intel.com
Wed Dec 22 11:47:21 CST 2010

I must say that I think that keeping the implementation requirement strong is critical at this phase.  The current standard is complex and implementation issues are likely to be subtle.  I don't think we have been through a single meeting where any one person has been "right" about their remembrance of the details of the standard.  If an implementation is such a burden that it is an impediment to getting something in, that raises a lot of questions about whether it should get in.  After all, we presumably want implementations to include these features in the foreseeable future, right?

That said, I think "unanimous" is probably slight overkill.  It would allow a single individual a procedural maneuver to kill (or substantially delay) minor changes to the standard or to a proposal.  I also think we will have to look at what qualifies as an "implementation" on a case by case basis.


From: mpi-forum-bounces at lists.mpi-forum.org [mpi-forum-bounces at lists.mpi-forum.org] On Behalf Of Graham, Richard L. [rlgraham at ornl.gov]
Sent: Wednesday, December 22, 2010 10:32 AM
To: Main MPI Forum mailing list
Subject: Re: [Mpi-forum] Proposal for the MPI-3 standardization process

I will take this, and what Rolf sent, and add to what we had decided at the meeting.  I think the main point of disagreement is on the implementation.  There was a very strong sense the an implementation is required before formal action by the forum.  However, since this is intended to  help the process, and not just create work, the notion was that the forum could decide to bypass this, but it would have to be a unanimous decision.


On 12/22/10 12:04 PM, "William Gropp" <wgropp at illinois.edu> wrote:

I have taken Rolf's version and further edited it to clarify the
language a bit and add the definition of simple majority that we have
always used.


mpi-forum mailing list
mpi-forum at lists.mpi-forum.org

More information about the mpi-forum mailing list