Skip to content
Howard Pritchard edited this page Feb 1, 2021 · 2 revisions

#01/21/20 webex notes

Attending: Howard Pritchard, Martin Schreiber, Martin Schulz

Agenda

  • MPI://MPI_COMM_PARENT process set discussion
  • Continue to discuss bubbles

MPI_COMM_PARENT process set

Dan's email from 1/22/21 was discussed. The consensus was that this seems somewhat like a late addition to address the lack of support for MPI_COMM_SPAWN in the context of sessions - at least not fully. The problem with the proposal is it seems to break some of the semantics for group and communicator construction based on the existing Sessions model.

Martin Schulz thought we should consider a more fully developed alternative to spawn for Sessions 2. Howard was of the same opinion.

Bubbles

Since Dan wasn't present, we decided to defer this conversation till later.

Operational issues discussion

Since Dan, Howard, and Martin Schulz are all very busy with the MPI standard this month, and Dan won't be able to resume till March, we decided to cancel further Sessions WG meetings in February, and resume March 1st.

Clone this wiki locally