Stan: Avoid terminal colors in messages

[andys8]

Solves #3089

[michaelpj]

Build fails with 8.10.7, I think because that forces an older version of stan that doesn't expose unCategory?

[andys8]

I'll have a look 🤔 Btw. why is hls-stan-plugin listed in stack-lts16.yaml but missing in stack-lts19.yaml and stack.yaml?

[uhbif19]

@andys8 Stan does not work with GHC >8.10 (kowainik/stan#423).

I does not understand how it seems to work with 9.2 in your case.

[andys8]

Maybe it's just skipped in that case?

https://github.com/andys8/haskell-language-server/blob/f274919bccc02d139295dedfc5f2e4ec8db313fd/plugins/hls-stan-plugin/hls-stan-plugin.cabal#L26-L29

[michaelpj]

Yeah, I think that means it was just unconditionally broken. There is a small mystery in that your original report said you were using GHC 9.2 which shouldn't have been possible!

[andys8]

There is a small mystery in that your original report said you were using GHC 9.2 which shouldn't have been possible!

Ah, I can help solve the mystery. The system ghc version is indeed 9.2 and hls was built with this version.

The project where the screenshot of the issue was taken from is a stack project using 8.10.7 (resolver: lts-18.23) though.

Still the output of --probe-tools in that project's directory looks like this. It also says "found on $PATH". When I created the issue, I wasn't aware the plugin is depending on the ghc version.

haskell-language-server-wrapper --probe-tools
haskell-language-server version: 1.7.0.0 (GHC: 9.2.4) (PATH: /home/andreas/.ghcup/bin/haskell-language-server-wrapper-1.7.0.1)
Tool versions found on the $PATH
cabal:          3.8.1.0
stack:          2.7.5
ghc:            9.2.4

[michaelpj]

Hmm, interesting. So for stack users this is potentially misleading, we really want to report the GHC that stack would use for the project or something.

[andys8]

Is the failing benchmark check related to this pull request?

And as a followup to the "ghc mystery" we could add the output of stack ghc -- --numeric-version to the output of --probe-tools, see #3093

[pepeiborra]

The bench failure is unrelated - our CI fails to upload the cache artifacts! I don't know why this happens, retried and hopefully it will go through but long term it would be great if someone could figure it out