Skip to content
Peter Scheibel edited this page Jan 5, 2022 · 10 revisions

Unlike many prior meetings, this particular meeting doesn't have any preset agenda items (so it will primarily consist of whatever questions attendees want to bring up).

Attendees

  • Peter Scheibel
  • Massimiliano Culpo
  • Richarda Butler
  • Srinath
  • Tammy Dahlgren
  • Phil Sakievich
  • Andrew Elble

Discussion

  • Srinath: status of ARM arch testing on AWS?
  • Phil: new failures with Clingo
    • Externals
    • Issue: 28201
    • Building packages with Spack, and having users maintain other Spack instances which refer to those installations as externals
    • Do upstreams work (e.g. in combination with reuse)?
      • Phil: not ideal for their case
        • At least once per month, create a new snapshot
        • Would need to manage one Spack instance per snapshot if using upstreams
  • Andrew: interested in status of separate concretization of build deps
    • Greg: compilers-as-dependencies expected this quarter
  • Massimiliano
    • Release schedule for 0.16
    • How long do we plan to support each major release?

Possible agenda items

  • Multiple providers of same virtual package
    • Case 1: we want blas from X and lapack from Y
    • Case 2: we actually want several instances of the same virtual (e.g. to run different instances of MPI for the same root)
  • Possibly revisit: https://github.com/spack/spack/discussions/24966
    • Concerning improvements to spack develop
  • Harmen (not sure if I can join): there are packages like libblastrampoline / libmpitrampoline which provide a blas / lapack / mpi interface to link to, and forwards calls to an actual blas / lapack / mpi provider lib. E.g. julia uses it to link to a blas interface for its binary deps, allowing the user to switch blas provider at runtime, avoiding abi issues. The problem is Spack only allows one provider per dag, but these packages both provides the virtual and depends on the virtual. How do we deal with this "composition" type of pattern? I was thinking: maybe we can relax unique provider per dag to unique provider per subgraph connected through link type deps? Then libblastrampoline can provide blas, and depend on blas as a run-type dependency.
Clone this wiki locally