FTS
Could it conditionally depend on [minad/writer-cps-transformers](https://github.com/minad/writer-cps-transformers)?
👍🏻 ~I believe the reason of performance improved here is `File.stream!/3` using [raw mode](https://hexdocs.pm/elixir/1.14/File.html#stream!/3-raw-files)~
it's possible reference ruby's implementation? https://github.com/rubyide/vscode-ruby/blob/7b5b6025caa2ebc140192e9add094aa34e7cec2d/packages/vscode-ruby-client/src/providers/highlight.ts
https://twitter.com/monadiccheng/status/1539583255317446658 by @TerrorJack > haskell-language-server with --nonmoving-gc is a lot smoother, when heap size goes beyond 10GiB
According to the benchmark from [scheduler](https://github.com/lehins/haskell-scheduler#benchmarks), `unliftio`'s [pooledMapConcurrently](https://hackage.haskell.org/package/unliftio-0.2.17/docs/UnliftIO-Async.html#v:pooledForConcurrently_)'s quite better than `async`'s `mapConcurrently`, maybe we could give it a try
https://github.com/Homebrew/homebrew-core/blob/master/Formula/c/cabal-install.rb Brew using official bindist to build from source.
May give [cabal2bazel](https://github.com/google/cabal2bazel) and [rules_haskell](https://haskell.build/) a try?
> I don't really like the name pop. In my mind, that always refers to an operation on a stack, or on one end of another sequence type. Just to...
`base` will include `nubOrd` and `nubOrdBy` in `Data.List` and `Data.List.NonEmpty` starting from version 4.23.0.0: https://github.com/haskell/core-libraries-committee/issues/336, https://github.com/ghc/ghc/commit/e4f6627