Create 2023-04-13.md

Author: david-christiansen

meetings/2023-04-13.md

@@ -0,0 +1,73 @@
+# TWG 2023-04-13
+
+[Last meeting's notes](https://github.com/haskellfoundation/tech-proposals/blob/main/meetings/2023-03-09.md)
+
+## Present
+ * David Thrane Christiansen
+ * Laurent P. René de Cotret
+ * Noon van der Silk
+ * Gershom Bazerman
+ * John Ericson
+ * Davean
+ * Jack Kensik
+ 
+
+## Updates on Projects
+
+### Error Message Index
+
+ * The site now deploys to both Github Pages and to Haskell.org hosting
+ * David will ping Davean to get stats enabled and to update DNS
+ * Summer of Haskell has proposals to do this for Cabal, David should follow up with feedback
+
+### Advisory DB
+
+ * The committee has been assembled and we're awaiting meeting number 1
+
+## [Standard Library Reform](https://github.com/haskellfoundation/tech-proposals/pull/47)
+
+ * David: it seems like there's broad support on the [CLC thread](https://github.com/haskell/core-libraries-committee/issues/145), but that [Bodigrim has some concerns](https://github.com/haskell/core-libraries-committee/issues/145#issuecomment-1484182300). My sense is that his criticisms of the text are legitimate, and that the text is missing some important background information.
+ * Gershom: It seems it would be really useful for someone else to reformulate and categorize his concerns so that we can make sure to address them, as the comments in the thread are a bit interlocking and complicated.
+ * Davean: The CLC does own all the pieces if we break something, and the fear that moving some parts of `base` into a `ghc-base` that they don't scrutinize will lead to difficulties is a real and legitimate one.
+ * Gershom: Can we build a division of labor that leads to there being less, rather than more, work for CLC? That would be a good way to address these concerns.
+ * Gershom: The other issue seems to be the direction. Why split out the bottom of base, rather than putting something on top? We need a technical answer. The answer is that `base` becomes reinstallable (at least in principle), and that becomes more useful as time goes by. The text of the proposal doesn't motivate this as well, and the motivating language of the proposal is a bit loose (not all motivations are addressed, and not all changes are motivated).
+ * John: This seems like a fair summary. Ideally, we move things to `ghc-base`, but they return to `base` as quickly as possible. One way to address this is to have a mechanical means of checking that `base`'s exports don't break.
+ * Davean: Do we have any tools for this? There's a few tools for detecting PVP violations, but they haven't been maintained for years. See `mtl` --- it's reinstallable, but there was nonetheless a controversy around it.
+ * John: That would also be a good HF project.
+ * Davean: History shows that people don't want these.
+ * John: They break for software eng. regions.
+ * Gershom: A concern seems to be that if the GHC team changes `ghc-base` in a way that changes `base` due to re-exports, then CLC will be left holding the bag. One way to address this is to get an agreement that changes to `ghc-base` get a quick approval from CLC that they don't change the `base` API (not a committee decision per se, just "this doesn't break it, go for it")
+ * Davean: Today we can see the commit to `base`
+ * John: We're not proposing moving `base` to a new repo (though I'd like to someday), so this doesn't change the visiblity of changes.
+ * David: The effort to separate base can be disputed if it will help maintence effort for compiler development.
+ * Noon: Could we reasonably expect the CLC to pass this motion without Bodigrim's approval?
+ * David: We want to include Bodigrim's opinions. It's best to address everyone's concerns (as much as possible).
+* Davean: The problems in industry don't run into decoupeling concerns, but it may still be worth tryinf
+* David: I have seen some places this could have been useful
+* Davean: The changes are seemingly trivial, sometimes it can just be changed by SED , it is almost never the `base`
+* David: The problems seems to arise when new major versions of `base` only work with new major versions of transitive dependencies, eventually resulting in big changes required to a codebase to update. Haskell libraries typically don't do backports, so dependencies with PVP bounds may not trigger updates in the same way. Having a minor release of base would help incremental changes
+* Davean: 
+* John: social issues are difficult to solve with technical advances
+* Gershom: John wrote a great proposal which has a great minimum piece which cannot be both everything and minimal, any future will need to start with something
+* Davean: Someone should work on this project, but it might not be a good starting point. The thesis may not scale to the full proposal.
+* Gershom: This may help us get Bodigrim on board. We should clarify what we're asking from CLC. We don't need them to be cheerleaders or to love it or to think it's the most useful thing; we need to get the proposal in a state where they don't think it will hurt things and won't block it.
+* Davean: The biggest vote is the vote of effort. Nobody should stop people who want to do work on things.
+* John: There might be a political concern that we are trying to resolve a conflict between GHC and CLC in a GHC-favorable way.
+* Davean: I don't see it that way
+* John: It's not that way, but it could look that way.
+* Gershom: If we had multiple compilers like in the good old days (or multiple GHC backends, which seems more realistic) then we can have a `base` that works for them all, and this is a step in that direction.
+* John: This seems to be clearly the better architecture, and you either believe that good architecture leads to good use-value outcomes or you don't. A small document won't sway that.
+* Gershom: What's the plan of action?
+* John: The proposal needs updated, we need to talk to Bodigrim, and then we can hopefully vote on it. I was supposed to update this a month ago, I had a meeting with David after the last committee meeting and took good notes but didn't yet do the update. The meeting with Laurent where we sat down and actually edited was very useful.
+* Gershom: Davean is the best voice to talk to Bodigrim because he's most skeptical here. Could Davean, John, and Laurent as a mediator/facilitator have a meeting with Bodigrim, and then get together afterwards to fix it up?
+* There seems to be broad consensus that this is how to do it.
+
+## Other
+
+* The June meeting collides with Zurihac . We will discuss rescheduling in May.
+* Can "we" (the Haskell community) do something to showcase _nice_ relationships between CLC/GHC/big language changes/anything, to make people feel good about compatibility?
+
+## Summer of Haskell
+
+Gershom appealed to the committee to help fundraise for Summer of Haskell (as individuals or as companies).
+Fundraising in the current economic climate is difficult.