[css-scoping] [selectors] <compound-selector-list> should somehow affect :is() / :where() parsing

[emilio]

https://drafts.csswg.org/css-scoping/#host-selector takes a <compound-selector-list> (that is, doesn't accept combinators). This is good, as otherwise it'd break the shadow tree encapsulation.

However strictly per the syntax technically something like :host(:is(div div)) should parse. I think this is a spec bug.

[emilio]

Maybe :is() and :where() should be more subtly defined. I'd expect ::part(foo):is(:hover,:active) to parse, just like the individual selectors would, but I don't think ::part(foo):is([attr="value"]) should parse.

[emilio]

This may also be an issue for pseudo-elements, so ::pseudo:is(:hover) may be fine, but ::pseudo:is(p) doesn't make sense.

[andruud]

Also, should ::pseudo:is() be allowed? (I.e. empty :is() after pseudo elements).

I think probably yes (but matches nothing).

[Loirooriol]

@andruud I think that's implied by #3264. ::pseudo:is(:hover, :bogus) should be treated as ::pseudo:is(:hover), ::pseudo:is(:bogus) should be treated as ::pseudo:is(), and ::pseudo:is() should match nothing.

[tabatkins]

@fantasai checked in an edit to fix this:

The [=logical combination pseudo-classes=]
are allowed anywhere that any other [=pseudo-classes=] are allowed,
but pass any restrictions to their arguments.
(For example, if only [=compound selectors=] are allowed,
then only [=compound selectors=] are valid within an '':is()''.)

Just need CSSWG thumbs-up for the edit.

[css-meeting-bot]

The CSS Working Group just discussed [css-scoping] [selectors] <compound-selector-list> should somehow affect :is() / :where() parsing, and agreed to the following:

• RESOLVED: Accept the edit

The full IRC log of that discussion

<dael> TabAtkins: Summary is after some discussion across a few related issues, basic idea is logical combo pseudos that combine things together passes down any restrictions the outer pseudo is in
<fantasai> :is(), :where(), and :not() are the set in question
<dael> TabAtkins: You can use is in a compund context, but it has to be compound too. That way you can't smuggle in a more complicated selector
<dael> TabAtkins: Have spec text. Need approval
<TabAtkins> also :nth-child() i think
<dael> astearns: Text is in last comment. seems good to me
<dael> astearns: Other opinions?
<dael> astearns: Prop: Accept this restriction and the edit that defines it
<dael> fantasai: Kind of more of an expansion. Previously only allowed some things and adds :is and :where
<dael> TabAtkins: Lot of places where they were allowed but could contain whatever. But doesn't matter
<dael> astearns: Prop: Accept the edit
<dael> astearns: Obj?
<dael> RESOLVED: Accept the edit