Define the autofill pseudo-class.

[emilio]

• At least two implementers are interested (and none opposed):

  • Gecko (proposed it in :autofill pseudo-class. #6181)
  • Blink (:autofill pseudo-class. #6181 (comment))

• Tests are written and can be reviewed and commented upon at:

  • Add a test for :autofill. web-platform-tests/wpt#27098

• Implementation bugs are filed:

  • Chrome: https://bugs.chromium.org/p/chromium/issues/detail?id=1164311
  • Firefox: https://bugzilla.mozilla.org/show_bug.cgi?id=1685675
  • Safari: https://bugs.webkit.org/show_bug.cgi?id=220459

(See WHATWG Working Mode: Changes for more details.)

---

/semantics-other.html ( diff )

[annevk]

Why is the alias a may? Would any browser not implement it? Is there a test for the alias being done at parse time? And is there a more formal way of defining that?

[emilio]

It seemed a bit weird to require a WebKit prefix in the HTML spec, it was mostly given as a hint. The alias is somewhat needed for compat right now, but if authors adopt the unprefixed pseudo-class (once implemented everywhere), hopefully eventually it will not.

[annevk]

If everyone implements it we should require it. We do the same webkitMatchesSelector() in DOM for instance. And then if implementations decide to remove it we can do so and change the pass condition for the test, though I'm not sure if it's worth trying to remove these.

[domenic]

Strong +1 for requiring :-webkit-autofill. Even if CSS itself makes it optional, it should be required for HTML user agents.

[emilio]

Alright, changed it to must, will update the wpt PR as well.

[domenic]

Let's just do two <dt>s, one for each selector, so that they're on equal footing?

Also, is there any normative connection we can state with https://html.spec.whatwg.org/#autofill ? It seems like that section is all about the autocomplete="" processing model, so if user agents autofill values beyond just those annotated, then maybe there is no such connection...

[emilio]

Let's just do two

s, one for each selector, so that they're on equal footing?

Sure, done.

Also, is there any normative connection we can state with https://html.spec.whatwg.org/#autofill ? It seems like that section is all about the autocomplete="" processing model, so if user agents autofill values beyond just those annotated, then maybe there is no such connection...

Yeah, pretty sure autocomplete is not the only thing that is looked at for form autofill...

[domenic]

I pushed another commit that made them on more equal footing, and also added a note about other sources of autofilling. PTAL to make sure I haven't done anything wrong :).

[emilio]

Looks good, thanks! I'd keep the mention that :-webkit-autofill is legacy, but it looks good anyways :)