lint: support stylelint?

[MLoughry]

I see that the @griffel/babel-preset depends on @linaria, which in turn has mechanisms to run stylelint over the resulting CSS. However, I cannot seem to get it working with Griffel.

Is there already an undocumented way of doing this that could be documented? Or would it be possible to create a supported way to do this? We already have a number of stylelint rules that we would like to re-use after a migration without having to rewrite them as questionable ESLint rules.

[layershifter]

I see that the @griffel/babel-preset depends on @linaria, which in turn has mechanisms to run stylelint over the resulting CSS. However, I cannot seem to get it working with Griffel.

Currently we use just a part of Linaria that performs module evaluation, in background we are trying to move to v4 (#414), but there are still blockers that prevent us from its adoption 😿

Fair note: we actively collaborate with Linaria maintainer to resolve them.

Is there already an undocumented way of doing this that could be documented? Or would it be possible to create a supported way to do this?

There is no way to do this currently, however @ling1726 works on enhanced ESLint rule/plugin that runs against compiled CSS. The benefit of this approach that we can emit errors on source code and don't require build for a feedback loop that makes DX significantly better 💪

I think that it would be great to embed stylelint to these changes. We don't have it in opensource yet, but I will connect you in offline.

[ling1726]

Needs some investigation to see if we can implement a stylelint preprocessor for Griffel and get good source mapping support. It's essentially all that Linaria's stylelint package does https://github.com/callstack/linaria/blob/master/packages/stylelint/src/preprocessor.ts

The current work with eslint, does much of the same as a stylelint preprocessor would require - we'd just need to see if we need to do that work differently with stylelint

[NotWoods]

This is somewhat high priority for Loop. We really need linting for our CSS because our developers are using various too-new CSS features and not testing in older browsers nor testing for all themes we support.