Skip to content

Latest commit

 

History

History
54 lines (40 loc) · 3.83 KB

2023-10-20.md

File metadata and controls

54 lines (40 loc) · 3.83 KB
Raw

TWG 2023-10-20

Prior meeting's notes

Present

  • Bodigrim
  • John Ericson
  • Noon van der Silk
  • Gershom Bazerman
  • Laurent René de Cotret

Agenda

Haskell Foundation Technical Proposal #57

#57

  • Noon: hoping for some integration to motivate the funding of the second round. This is already a moderately long project; Noon will go with the consensus. There is a reticence to increasing the scope due to the monetary penalty of not achieving the milestones. Are there operational guidelines regarding the way the funding is disbursed?
  • Gershom: Excellent question. Funding for a volunteer has never been approved by the TWG.
  • Bodigrim: If this project gets funded, we expect to specify 3 milestones, each with an expected duration of 1 month. Better to limit the scope of this project to 3 months. The board will decide on the specifics.
  • Gershom: OK with a 3-month rolling contract but details out of the TWG hands.
  • Noon: Do you think we could ask for an in-kind audit of the Haskell code written by the proposer, at the end of the project?
  • Gershom: hopes that the companies that can perform this audit, would be users of the new library. Formal audits are difficult but similar work will be performed by engineers employed by companies using this new library. Gershom looked into the proposer and stakeholders who are all showing interest. We could encourage stakeholders to help improve the library.
  • Bodigrim: no current in-kind auditing resources at the HF.
  • John: Ask seems a bit high for a library which is only bindings.
  • Gershom and Bodigrim: Bindings are complex, and the proposal will also result in a high-level library which must be designed.
  • Gershom: multiplatform CI of Haskell/C++ bindings is a lot of work in itself.

Voting proceeded with a unanimous approval (5 yes) of the proposal.

Haskell Foundation Technical Proposal #56

#57

  • John plans a revision of the document as follows:
    • Splitting the AST out of the ghc package is no longer part of the proposal, but will be contributed by John as a volunteer;
    • Haddock use of the split-out AST will be moved to a Future Work section;
    • Splitting out the AST out of the ghc package will demonstrate we can have a second library without adversely impacting GHC development
    • HF-funded part 2 -- splitting out the GHC parser into its own package -- would still benefit from the AST being a standalone package, reducing risk.
  • John: Unclear about the Haddock GSoC status and how to salvage it
  • Noon: How likely is it that splitting the AST into its own package will be merged in a timely manner? How interruptive is this splitting of the AST?
  • John: Intent is to get approval beforehand from GHC-HQ.
  • Gershom: Hopefully the AST isn’t touched a lot;
  • John: Work on GHC mostly happens on the typechecker.
  • Gershom: GHC-HQ pre-approval is most important; TWG approval is mostly symbolic until the second part (splitting out the parser).
  • Laurent: will move the Haddock content to a Future Work section

Other business

Incoming HFTP regarding Cabal changes to the ‘Custom’ build type. This work will need to be approved by the Cabal maintainers; the TWG would only act as a forum where all stakeholders may discuss.

  • Gershom: make sure to look into security ramifications of the resulting proposal, which is always a concern with hooks-type software extension points. Might consider authorization ‘effects’ for Setup.hs files.
  • John: Good idea to separate the generation of a build plan (possible using hooks), and executing the build plan in a non-extended part