Hello all
We will be having our first teleconference of the year on Thursday Jan 8th,
at 11am US Eastern. As usual, the connection info is available from the recaptcha at https://openpmix.org/captcha/ This month's telecon will deal with the following topics:
* brief recap of the two revised roadmap meetings that occurred in Dec. No conclusions were reached, but I think a more general understanding of the problems we face was achieved. There is an issue tracking some thoughts about versioning: https://github.com/openpmix/openpmix/issues/3733. Feel free to contribute. As stated in that issue, I’m currently targeting a PMIx 6.1 and PRRTE v4.1 series, probably releasing in late Jan - in lieu of any further outcomes from these discussions.
* I have posted release candidates for PMIx v5.0.10 and PRRTE v3.0.13. Please give them a test if you can. Targeting release for mid-Jan, depending on feedback
* discussion of future directions. PMIx is pretty stable at this point - I don’t expect anything more than some added attributes and cleanup/completion of the Python bindings. PRRTE is undergoing some significant change for v4.1, mostly in support for multi-app operations. Looking further ahead, a few of us are starting work on extending PRRTE to support truly elastic operations
PMIx commits
* cleanup of debugger operations
* fix cmd line option checker
* few fixes to the bitmap class
* fix refresh cache request
* cleanup of GitHub code check results
PRRTE commits
* fix prte_info to output correct version string
* multiple mapper fixes
* some cleanup on hostfile parsing
* shift rmaps lsf support into its own component
* allow setting of all five major cmd line options (map, rank, bind, display, runtime-options) by MCA param
* add early support for multi-app map-by directives
* cleanup bind output when partial allocations exist
* improve auto-detection of hetero nodes
* improve support for asymmetric topologies