webkit-prefixed pseudo-elements are always treated valid in WebKit and Blink

[miketaylr]

The data is now showing ~40% of page loads have unknown -webkit pseudo elements from the Chrome use counter, which is a significant proportion.

w3c/csswg-drafts#2156 (comment)

[miketaylr]

See https://bugzilla.mozilla.org/show_bug.cgi?id=1424106 and https://bugzilla.mozilla.org/show_bug.cgi?id=1331793

[upsuper]

Note for the spec: Blink seems to lowercase the pseudo-element string. This behavior should probably also be put into the spec if we going to do that.

[annevk]

For the purposes of selectorText you mean? Does it only do that for prefixed pseudo-elements or all of them?

@fantasai @tabatkins it seems we should put this "always treat webkit-prefixed pseudo-elements as valid" straight into the Selectors specification, no?

[upsuper]

FWIW, several a new web-platform test for this is being added in web-platform-tests/wpt#12673.

[miketaylr]

it seems we should put this "always treat webkit-prefixed pseudo-elements as valid" straight into the Selectors specification, no?

+1, if possible.

[tabatkins]

Sigh, yeah.

[tabatkins]

https://drafts.csswg.org/selectors-4/#compat

[annevk]

@tabatkins cool, thanks. Where is serializing of selectors defined?

[tabatkins]

https://drafts.csswg.org/cssom/#serializing-selectors

[miketaylr]

Assuming that sticks, looks like we can close here. Thanks everyone!

[annevk]

I think the serializing definition needs to be updated as these prefixed pseudo-elements get lowercased (not defined).

I also wonder if since they are not technically psuedo-elements some other language needs to be added, but maybe it's okay.

[tabatkins]

I think the serializing definition needs to be updated as these prefixed pseudo-elements get lowercased (not defined).

CSSOM needs to define in general that pseudo-class/element names are lowercased; these specific pseudos aren't special in that regard. (See http://software.hixie.ch/utilities/js/live-dom-viewer/saved/6160.) So that's an issue on OM, but nothing that should stop this specific issue from closing.

I also wonder if since they are not technically psuedo-elements some other language needs to be added, but maybe it's okay.

We don't care what they are in the tree (which is "nothing"), we care about what they are in the context of Selectors, which is definitely "pseudo-element selectors". No new terminology needs to be added.