Meeting 2019 04 26

        Josh Hursey (IBM)
	Kathryn Mohror (LLNL)
	David Solt (IBM)
	Swaroop Pophale (ORNL)
	Jim Garlick (LLNL)
	Swen Boehm (ORNL)
	Thomas Naughton (ORNL)
	Stephen Herbein (LLNL)

Agenda and Notes

• Continue discussion on PMIx standardization process
  • Discuss: https://github.com/pmix/pmix-standard/issues/181
    • Silence is lack of dissent - has the issue that you don’t know how many people actually looked at the item.
      • We can document who attended the teleconf
      • Then for those that don’t attend the teleconf they 👍 on the GitHub PR
      • Take a straw poll at the reading - keep poll window open for 24 hours (more? Whole week?).
        • If there is a dissent then it can be documented on the ticket.
        • Use emoji on the comment to track positive/negative straw poll votes
      • That keeps it light weight, but gives a sense of how many people viewed the ticket/PR
      • Second teleconf - Don’t use “Voting” use “Acceptance into standard”
      • If folks need more time to review they should post on the ticket to request more time to delay Acceptance and merging.
    • Move to take Issue 181 into a PR change to README
  • Discuss: https://github.com/pmix/pmix-standard/issues/179
    • PR “Accepted” and labeled as “experimental”
      • “Experimental” is somewhat negative. Is there a better term we can use to express that it’s ‘new’ - maybe colors, or different terms, ...
    • “experimental” to “widely used”
      • 2+ users of the interface. How do we define “users”
        • Large following of a single user library
        • Multiple use cases can move the label
        • Multiple product consumers
      • Having someone else use the interface to verify it’s stability
    • “Widely used” to “stable”
      • Time based or implementation based
    • Alternatively to a measurable means of transition, maybe we can just have a solid argument. Objective vs subjective.
    • Moving between labels happens in a PR and goes through the normal acceptance process.
    • Let’s keep this discussion going on the ticket
  • We need a scoping statement to determine if something belongs in PMIx or not
  • Try to limit this to first ½ of the meeting
• (Dave) Working group: Implementation agnostic document
  • https://github.com/pmix/pmix-standard/issues/175
  • Should we remove the concept of the PMIx Reference Implementation (PRI) from the document, or make more of an attempt to isolate that into specific ‘advice’ sections.
  • Distinction between server interface and RM functionality
    • Could we better isolate the client and server, so it’s possible for a PMIx compatible library not support the server side interfaces.
    • A PMIx implementation can opt-into the server side interfaces
    • Server-side interfaces still part of the standard to stabilize that aspect of the existing PRI.
      • This might need some more generalization to work with different RM implementation models
      • Could we slice along a wire protocol for client-server messaging
    • Do we expect a different server side interface?
  • 3 concepts - maybe tied to document divisions (chapters, actual documents, …):
    • Client side API
    • Wire protocol
    • Server side API
  • Next step - propose another meeting to define the high level objectives, and work to do (review/rework/write)
• (Stephen) Working group: Slicing/Grouping of functionality
  • Some overlap with the Implementation agnostic document working group
  • Will open an issue to discuss how to slice the document
• (Josh) Use case gathering
  • https://github.com/pmix/pmix-standard/wiki#use-cases

Action Items

• (Josh) Move notes to PMIx Standard wiki
• (Josh) Create a PR form of Issue 181 for reading.
• (Dave) Send out a doodle poll for Implementation agnostic document side meeting
  • Send email to the mailing list
• (Stephen) Open an issue specific to Slicing/Grouping of functionality
• (Josh) Sketch a template issue for use cases with a single example
  • Open discussion on suggest template for these use case issues
  • Create a few Use Case Issues on GH to experiment with the template and set some examples for new participants.