Rule idea selector-pseudo-class-no-empty-parentheses

[alexander-akait]

Describe the issue. Is it a bug or a feature request (new rule, new option, etc.)?

Maybe

Invalid

input:not() { }

:matches() { }

Valid

input:not([type="submit"]) { }

:matches(section, article, aside, nav) h1 { }

Include nth-child, nth-last-child and etc.

Which rule, if any, is this issue related to?

selector-pseudo-class-no-empty-selector

What CSS is needed to reproduce this issue?

no matter

What stylelint configuration is needed to reproduce this issue?

e.g.

{
  "rules": {
    "selector-pseudo-class-no-empty-selector": true
  }
}

Which version of stylelint are you using?

no matter

How are you running stylelint: CLI, PostCSS plugin, Node API?

no matter

Does your issue relate to non-standard syntax (e.g. SCSS, nesting, etc.)?

no matter

What did you expect to happen?

no matter

What actually happened (e.g. what warnings or errors you are getting)?

no matter

[jeddy3]

I think we might have taken a wrong turn when we started using these *-no-empty rules as proxies for testing the validity of things.

Previous *-no-empty rules (e.g. block-no-empty, no-empty-source and comment-no-empty) allowed a user to specify a preference i.e. an empty block, comment or source is still valid CSS. selector-no-empty was the first *-no-empty rule to break with this pattern.

I think we're creeping into spec parser territory here, which I think is better handled in projects like csstree and css-values and then wrapped as stylelint plugins. These plugins are only applicable to users who write spec-based CSS i.e. those who do not use language extensions introduced by PostCSS plugins or preprocessors. I think such plugins would be more powerful than these piecemeal attempts to check for validity.

So, my suggestion is we close this issue in favour of someone creating a selector validator.

I also suggest we undo our mistake with selector-no-empty and deprecate that rule in 8.0.0. Along with media-feature-no-missing-punctuation as that is a proxy for validity too. Arguably the
*-no-unknown rules belong in a plugin pack too.

Thoughts?

[ntwb]

I think you're right @jeddy3, stylelint is not a CSS spec parser 👍

[alexander-akait]

@jeddy3 agree, also i agree selector-no-empty should be deprecated.

[davidtheclark]

👍 from me, too.