sheaf

It would be good to have a smaller reproducer. I tried the following but it seems to work (at least with #90): ```hs k1 :: (((m6 + k1) + n14)...

> `git bisect` says: [2d07efe](https://github.com/haskell/cabal/commit/2d07efe3580a9fe8f1191d26056307d47205462c) [#9969](https://github.com/haskell/cabal/pull/9969) > > [@sheaf](https://github.com/sheaf) Thanks for bisecting this. I will try to figure out what is going on.

I can't reproduce this on NixOS. With cabal-install 3.14.2.0, the example works correctly for me. Here is the `ldd` output: ``` ldd dist-newstyle/build/x86_64-linux/ghc-9.4.6/hscurses-1.5.0.0/build/contact-manager/contact-manager linux-vdso.so.1 (0x00007fffbcfc7000) libm.so.6 => /nix/store/ld03l52xq2ssn4x0g5asypsxqls40497-glibc-2.37-8/lib/libm.so.6 (0x00007ffb00489000) libncursesw.so.6...

I investigated with the help of @mpickering. Here is our analysis: 1. `hscurses.cabal` contains `extra-libraries: ncurses`. 2. `hscurses` use the configure build type, with the configure script producing `hscurses.buildinfo`. The...

It seems to me that the plugin's behaviour is correct. The only issue is GHC not running typechecking plugins when doing pattern match warnings, which is a bug that I...

In type-checking plugins, it's common to want to rewrite type family applications. For the sake of illustration, let's say we want to implement the rewrite `F (G x) ~> x`,...

> Please don't bury documentation in a comment to an issue. Right, I was just in the middle of writing a question to ask what you think I should add...

Indeed, it can be a bit tricky to thread through coercions in order to write an evidence term for typeclass constraints; it's not as easy as for equality constraints. For...

> Or rather GHC would force us to do it in opposite order, doesn't it? Yes, that's right. > GHC Plugin API doesn't say whether we can introduce new wanteds...

> > I would have to test that. I imagine it's OK but I'm not 100% sure. > > That would be great. Then we can give better advice in...