2023 12 11 webex

Agenda

• Discuss feedback from https://github.com/mpi-forum/mpi-standard/pull/938 (the relax provenance of pset names)
• Discuss resync with FT group in CY 24

Discussion

For PR 938 consider whether to change anything? Maybe add some context info for the sentence at line 712. No one on the call today was wanting to add an advice to users/implementors about where this process set name could come from.

Dan says since first vote passed we should leave text as is for second vote.

Dan says the wording does lead to a possible question - Having returned that name once from call to MPI_SESSION_GET_NTH_PSET how long is it still valid? We think the answer to this is yes. Across various session handles should I expect a pset name to be valid when using a different session handle? We think the answer to this is not necesarily.

Martin Schreiber now thinks we do need to change the text some at line 712 to help with clarification of these questions.

Discuss when a "failure" would occur if processes in a process set no longer exist/are reachable.

Options

• go with what's in PR
• no-no vote addition

Action items

• Howard will add no-no commit addition to a cpy of PR 938 to be discussed next week