Thanks! Turns out there was an incorrect type annotation, and putting it behind a cpp fixed it!
Then I got a bunch of new instance errors, which I think I was able to fix them all in 016daf !
1 Like
We got the first successful CI build with ghc-9.10, once the project built the test worked! In summary:
- No changes were required in our codebase, though monocle is relatively simple so I hope it will keep on building with future compilers.
- About 20 dependencies (in nixpkgs) needed their upper bounds to be relaxed.
- A couple dependencies needed their tests to be disabled because doctest failed with the new x-partial warnings.
- morpheus-graphql, hlint and weeder are already compatible with ghc-9.10, but not on hackage, so we had to pull HEAD.
- proto3-suite is the only thing that needed to be modified, and it was painful (see PR#262).
I would say this went well, and I’m glad we waited; last time we upgraded to ghc-9.6 when it was fresh from the oven (for the record dot syntax extension) and that required much more work. I was about to do the same for ghc-9.12 (to get the named default extension), but without HLS it’s a no go.
Thank you everyone for maintaining the ecosystem!
2 Likes