[css‑shadow‑parts] What is the specificity of ::part(…) with multiple <ident>s?

[ExE-Boss]

https://drafts.csswg.org/css-shadow-parts/#part

Since the part name is similar to a class name, I would expect that ::part(active tab) would have a higher specificity than ::part(tab) or ::part(active).

---

See also: #1915, #3502, WICG/webcomponents#847

[tabatkins]

Sounds reasonable, but do we want this to interact class-like with real classes?

That is, do the following have the same specificity, so the result is solely determined by order?

.foo::part(tab) {color: blue;}
::part(active tab) {color: red;}

[emilio]

I think I'd rather keep it working as a regular pseudo-element selector than complicating the specificity rules, but that's just my opinion :)

[ExE-Boss]

Sounds reasonable, but do we want this to interact class-like with real classes?

That is, do the following have the same specificity, so the result is solely determined by order?

.foo::part(tab) {color: blue;}
::part(active tab) {color: red;}

Well, .foo selects the shadow host in the light tree, whereas ::part(active tab) selects the part in the shadow tree, so ¯\_(ツ)_/¯.

[emilio]

I don't think that matters? Both can end up matching the same element, so they need to be sorted by specificity and thus tab's question still stands.