Should open `dialog` elements match `:open`?

[josepharhar]

This was brought up here: #8221 (comment)

The new popover attribute adds the :open pseudo-selector which matches elements with the popover attribute when they are open and are in the top layer. It would make sense for dialog elements to also match this pseudo selector when they are open.

@mfreed7 @annevk

[annevk]

cc @whatwg/css

[tabatkins]

No strong opinion, but yeah, seems like it's an appropriate match for the pseudo, and authors would probably expect it. It's certainly "an element that has both open and closed states".

(And remember we have the :closed complementary pseudo-class!)

[tabatkins]

Oh, by "no strong opinion" I apparently mean "yes, I wrote that details was an exemplar element for the pseudo-class in the spec".

So yeah, please define details as matching :open/:closed. ^_^

[nt1m]

I think it's a good idea, given the open attribute doesn't necessarily reflect the state (for dialog specifically). We should probably also change the default UA styles too.

[domenic]

given the open attribute doesn't necessarily reflect the state (for dialog specifically)

I still think we could fix this. Someone just needs to sign up to drive #5802 to completion. (I don't think I have the bandwidth to be that person, unfortunately.) #5802 (comment) is a fairly concrete proposal, although were was some followup that might be worth resolving.

@josepharhar: given #9077, this can be closed?

[josepharhar]

@josepharhar: given #9077, this can be closed?

No, that PR is for the popover attribute, not dialog elements.
Dialog elements should still match :open and :closed as they are specified in CSS

[josepharhar]

So the pseudo-class for popover was renamed to :popover-open in order to differentiate it from :open since any element can be a popover and be open as a popover separately from a dialog, details, or select element being open.

We can and should still make :open and :closed apply to dialogs, details, and selects as tab mentioned is written in the selectors spec.

[annevk]

Well, someone would first have to define that mapping in HTML's Pseudo-classes section. The Selectors specification gives host-agnostic guidance so we still need to define exactly when non-generic things defined therein match.

cc @tabatkins