Skip to content

Latest commit

 

History

History
66 lines (52 loc) · 3.31 KB

2022-11-28.md

File metadata and controls

66 lines (52 loc) · 3.31 KB

SWG 2022-11-28

Meeting URL: https://meet.jit.si/StableHaskellMeetBiWeekly

Previous meeting notes

Agenda & Notes

New items

  • Any volunteers to run next meeting? Trevis will be on vacation
    • David to run
  • Also holiday/end of year
    • Skipping meeting that would have fallen on December 26th.
    • Trevis to cancel meeting invite

Updates on in-progress work

  • GHC API discussed previously

    • Previously holding the token: David
    • No progress
    • Holding the token: David

  • Update on Language/compiler features to help stability

    • Previously holding the token: David

    • Need to get a quote for some of this

    • David has meeting for us.

    • Asking for implementation of this

      • Of all the things suggested our our Language/compiler features to help stability page, this is the only one that is ready, and cheap-looking.
      • This already has an accepted proposal
      • Implementation tracked at #4879, although it needs to be updated to reflect the current state of the proposal

    • Simon and David to ask GHC devs for volunteers

    • Holding the token: David+Simon

  • Creating and maintaining a set of hlint rules to promote stability

    • Previously holding the token: Trevis
    • Submitting a feature request soon
    • Holding the token: Trevis

  • Bulletin discussed previously

    • Previously holding the token: Chris
    • No progress

  • GHC warning policy document as discussed previously

    • Previously holding the token: Chris
    • No progress

  • What is the stable API of base

    • CLC has said all of the exposed modules

    • Can we mark all of the modules with the stability haddock field?

    • Related: Namespace with Internal modules.

    • Zurihac project? Go through the 232 exposed modules and categorise them as

      • Should be exposed
      • Should not be exposed
      • Needs further attention
        • look at API function by function
        • identify the subset that should be exposed
        • maybe make new modules for these exposed ones

    • More generally, a package may well want to expose "internal" modules, say for testing frameworks; but those internal modules should not be subject to the PVP; and clients should not expect to able to rely on their stablility.

      • Perhaps having Internal in the name? This is a widely used convention. But then the PVP should say this. This idea is tracked as haskell/pvp#8
      • Ben will open a MR.
      • c.f. Opaleye conventions

Parked for expected absence of token holder