Meeting URL: https://meet.jit.si/StableHaskellMeetBiWeekly
Previous meeting notes
-
Transition for loopy-superclass-solve
- Is it reasonable to have 9.6.4 with it in Wcompat?
- What is the policy?
- https://github.com/ghc-proposals/ghc-proposals/blob/wip/general-rules/principles.rst#3ghc-stability-principles
- Says no warning stability
- Do we want
-Wcompat?- Suggestion is to removal; Simon will add to GHC proposal ghc-proposals/ghc-proposals#625
- https://github.com/ghc-proposals/ghc-proposals/blob/wip/general-rules/principles.rst#3ghc-stability-principles
- Future releases of 9.6/9.8 should include ability to name; Ben has marked !11421 for backport, which means it'll be in any future point releases of these compilers.
- Cannot happen again in the future by https://gitlab.haskell.org/ghc/ghc/-/merge_requests/11421
- How can prevent the documentation miss?
- In particular, which flags are in which flag-groups. See https://gitlab.haskell.org/ghc/ghc/-/issues/24255
-
Reinstallable base
- Somehow making Template Haskell more stable is potentially more impactful than reinstallable base
- See https://gitlab.haskell.org/ghc/ghc/-/issues/20828, https://gitlab.haskell.org/ghc/ghc/-/issues/24021. Some good discussion here!
- Relevant package: https://hackage.haskell.org/package/th-abstraction
- Might need more advertisement/promotion to use this over
template-haskellitself.
- Might need more advertisement/promotion to use this over
- Advertise a potential HF project to design/build a th-abstraction like project.
- Need to write what that would look like
- Goals/design issues.
- Need to write what that would look like
Thoughts from Ryan GL Scott on th-abstraction's design:
Briefly, my thoughts on th-abstraction is that it attempts to satisfy two related (but distinct) concerns:
- It provides a semantic, rather than a syntactic, representation of data type definitions that applies equally to both ordinary data types and data family instances.
- It provides an API that smoothes over the differences in the template-haskell API across GHC versions.
Because it is trying to do both things simultaneously, I'm not convinced that th-abstraction would be an ideal API for reification in general. Nevertheless, perhaps there are some lessons that could be learned from it going forward.
Call to action for Template Haskell: https://edit.smart-cactus.org/C-0XLaG1QoaEo8qhI6sK8w (Simon will start)
- None
-
Bulletin discussed previously
- Previously holding the token: Chris
-
GHC warning policy document as discussed previously
- Previously holding the token: Chris
-
Creating and maintaining a set of hlint rules to promote stability
- Previously holding the token: Trevis
- Opened issue asking if we can write rules for typeclasses.