1419323 - Fix custom element WPT failures at CustomElementRegistry.html

Status: RESOLVED DUPLICATE of bug 1462453

(Core :: DOM: Core & HTML, defect, P3)

Description

[John Dai[:jdai]]

It remains 3 test failures at CustomElementRegistry.html[1][2][3].

[1] https://searchfox.org/mozilla-central/rev/919dce54f43356c22d6ff6b81c07ef412b1bf933/testing/web-platform/tests/custom-elements/CustomElementRegistry.html
[2] https://searchfox.org/mozilla-central/rev/919dce54f43356c22d6ff6b81c07ef412b1bf933/testing/web-platform/meta/custom-elements/CustomElementRegistry.html.ini
[3] http://w3c-test.org/custom-elements/CustomElementRegistry.html

Comment 1

[Blake Kaplan (:mrbkap) (inactive)]

One of the failures here (the "customElements.define must get callbacks of the constructor prototype") is because the custom elements spec requires that we retrieve the lifetime callbacks in a specific order [1] (step 10.3-4). Internally, we use a WebIDL dictionary to store the lifetime callbacks; however, WebIDL dictionaries specify that their keys should be retrieved in alphabetical order. I can pretty write custom code to retrieve the lifetime callbacks in the custom order, but I was wondering if it would be worth changing the spec to retrieve the callbacks in alphabetical order (which would save us some code and also makes a bit more sense to me).

What do you think, Olli?

[1] https://w3c.github.io/webcomponents/spec/custom/#dom-customelementregistry-define

Comment 2

[Olli Pettay [:smaug][[e-mail]]]

oh, that sounds like a bug-ish in the spec. Better to follow the common web conventions, so WebIDL.

Please use https://html.spec.whatwg.org/#custom-elements as the spec.

Comment 3

[Blake Kaplan (:mrbkap) (inactive)]

https://github.com/whatwg/html/issues/3580

Comment 4

[Blake Kaplan (:mrbkap) (inactive)]

Almost all of the remaining failures in this test are due to the property order issue. There's one easy fix that's also needed and I'll attach it here next.

Comment 5

[Blake Kaplan (:mrbkap) (inactive)]

Attachment: Bug 1419323 - Remove some unnecessary code.

This code appears to be trying to verify that all of the callbacks exist and
are of the correct type, but our dictionary code does that for us.

The additional failing tests depend on the order that we retrieve the
callbacks in.

Review commit: https://reviewboard.mozilla.org/r/229560/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/229560/

Comment 6

[Blake Kaplan (:mrbkap) (inactive)]

Attachment: Fixing the test

This makes the test expect the callbacks to be retrieved in alphabetical order.

MozReview-Commit-ID: DjyjQVKwQOb

Comment 7

[Olli Pettay [:smaug][[e-mail]]]

Comment on attachment 8960821 [details]
Bug 1419323 - Remove some unnecessary code.

https://reviewboard.mozilla.org/r/229560/#review235396

::: commit-message-6d0f8:3
(Diff revision 1)
> +Bug 1419323 - Remove some unnecessary code. r=smaug
> +
> +This code appears to be trying to verify that all of the callbacks exist and

Not really. It does the checks only if value isn't undefined.

Comment 8

[Olli Pettay [:smaug][[e-mail]]]

I guess this is now sort of a dup of bug 1462453.