jade

~~the import quickfix seems broken on 9.4: https://github.com/haskell/haskell-language-server/issues/3195 I have a suspicion that the diagnostic messages were changed just enough to break our regexes, which makes that possibly: https://github.com/haskell/haskell-language-server/issues/2014~~ nope,...

Oh, there's another fun way to cause this bug: do this: `cabal.project`: ``` packages: . ``` `cabal.project.local`: ``` packages: . ``` Surprisingly, I've hit this multiple times by accident.

Hi! I'm currently writing a blog post on a workaround for the absence of this feature so I have lots of context on this. The core cabal-relevant issue here is...

Correct; this uhh weird emergent use of cabal, we'll call it, is described in this post: https://jade.fyi/blog/nix-hls-for-deps/ under the subheading "yolo method". I use this regularly at work to deal...

Hmmm. I thought that was impossible since it would have to apply to hackage2nix to add dependencies as it needs to, but I think the dependency could be added by...

Thanks for the advice. It's been dealt with in nixpkgs.

It wouldn't, surprisingly. Long story, marred by my own cursed workflows on top of a codebase that needs to be weird to be usable. The reason is that there's overlapping...

This is definitely an aarch64 bug; I reproduced it on my aarch64-darwin M1 mac using ghc942 from nixpkgs. https://gist.github.com/20536c5a14a5194200e61dc965d851a0

I've filed it upstream and sent it to some compiler folks Mercury is consulting with. This is what I would consider an extremely bad compiler bug. https://gitlab.haskell.org/ghc/ghc/-/issues/22282

Looking into this further, it looks like we could build a `hs-opentelemetry-instrumentation-servant-client` that implements RunClient, then transform the client monad.