Fix no-descending-specificity false positives for pseudo-classes
#6195
Conversation
| */ | ||
| function checkSelector(selectorNode, ruleNode, comparisonContext) { | ||
| const selector = selectorNode.toString(); | ||
| const referenceSelectorNode = lastCompoundSelectorWithoutPseudoClasses(selectorNode); | ||
| const referenceSelector = lastCompoundSelectorWithoutPseudoClasses(selectorNode); |
There was a problem hiding this comment.
[note] lastCompoundSelectorWithoutPseudoClasses() returns a string, not a node. So I think referenceSelector is a more readable name.
| ); | ||
| }); | ||
|
|
||
| if (nodesWithoutPseudoClasses.length === 0) return undefined; |
There was a problem hiding this comment.
[note] If nodesWithoutPseudoClasses is empty, the old version of this function returns an empty string (''). This makes it more difficult to handle irregular values. So the current code returns undefined instead.
|
I created another example, But I'm unsure whether it has anything to do with this problem. |
|
@richex-cn Thank you for the advice! I've added a test case via 423095a, and this fix works in that case. 👍🏼 |
see https://github.com/stylelint/stylelint/blob/main/docs/user-guide/rules/about.md#user-defined-ignore |
|
@Mouvedia Thanks for providing the guidance. Indeed, I think it may be better that built-in rules don't support non-standard syntaxes like CSS Modules for simplicity and maintainability. |
|
I don't have an opinion. I am just saying let's be consistent. |
|
Just getting back around to this. It's a can of worms and I don't think I've fully grokked it yet. I think we want to address functional pseudo-classes more generally rather than just the non-standard The rule currently considers the following comparable when it should not: a :where(a, b) {}
:where(c, d) {}I haven't dug into the implementation of the rule deeply, but I think we should be including the Then, as (We could add a different option which lets a user specify a list of functional pseudo-class to treat like |
Let's add an |
|
@jeddy3 Thanks for the suggestion. I mistakenly thought the CSS Modules specific code was required; indeed, it's unnecessary. See the commit 68bcaa which makes the test suite pass. If the |
jeddy3
changed the title
no-descending-specificity false positives for :global() pseudo-classno-descending-specificity false positives for pseudo-classes
Closes #4010
This pull request aims to fix false positives for the
no-descending-specificityrule about the CSS Modules:global()pseudo-class.I'm unfamiliar with CSS Modules, so I'm unsure whether this added code specific to
:global()is suitable. Perhaps, someone may know a more smart solution.In addition, I don't know if we should address the
:local()pseudo-class. Please let me know if someone knows anything about it.See also #4010 (comment) and the reproduction demo.