[font-loading] FontFaceSetLoadEvent's fontfaces should use FrozenArray

[foolip]

https://drafts.csswg.org/css-font-loading/#fontfacesetloadevent

Current IDL:

[Constructor(DOMString type, optional FontFaceSetLoadEventInit eventInitDict),
 Exposed=Window,Worker]
interface FontFaceSetLoadEvent : Event {
  readonly attribute sequence<FontFace> fontfaces;
};

https://heycam.github.io/webidl/#idl-sequence says "Sequences must not be used as the type of an attribute or constant" so this needs to be:

[Constructor(DOMString type, optional FontFaceSetLoadEventInit eventInitDict),
 Exposed=Window,Worker]
interface FontFaceSetLoadEvent : Event {
  [SameObject] readonly attribute FrozenArray<FontFace> fontfaces;
};

[SameObject] thrown in because it's true, a drive-by.

A test for this would have to check that event.fontfaces===event.fontfaces, and that new FontFaceSetLoadEvent('type', { fontfaces: x }).fontfaces!==x but that the array length and members are equal.

Lists of things are fun. @LoonyBean

[bzbarsky]

For what it's worth, Gecko's IDL for this is:

[Cached, Constant] readonly attribute sequence<FontFace> fontfaces;

which is just missing [Frozen] to be functionally equivalent to [SameObject] readonly attribute FrozenArray<FontFace> fontfaces;.

The [Frozen] thing came up during code review for the Gecko code (the original patch did include it) and was supposed to lead to the filing of a spec issue, because the spec didn't define the behavior of this getter at all. It looks like the spec issue never got filed. :(

[foolip]

Great! @LoonyBean did write a test for this that would be easy to put in web-platform-tests, but I guess it's still csswg-test for this spec?

[bzbarsky]

Are there web platform tests added for this?

[foolip]

@LoonyBean hasn't landed the Chromium change for this yet, but it has a test that would be easy to contribute if putting it in web-platform-tests is acceptable. There isn't even a font-loading directory in csswg-test. @gsnedders, please tell us what to do :)

[heycam]

Yeah, looks like I forgot to file an issue at the time. :( Although a few months later @RByers reported the issue (and I replied to it!), but I guess it never got acted upon. Anyway, thanks for getting it fixed now, @foolip.

[gsnedders]

@astearns do we have any policy on where new testsuites go?

[astearns]

We're moving everything to wpt, so if it can go into wpt to begin with I don't see any reason not to do that.

[foolip]

@LoonyBean, what happened in the end with the tests for this? Looks like Blink's test in third_party/WebKit/LayoutTests/fast/css/fontfacesetloadevent-constructor.html could easily be upstreamed if we just know where to put it? @gsnedders, maybe you can just ping this issue after csswg-tests are merged instead of us trying to figure out an interim place for the test?

[gsnedders]

@foolip can you file an issue for the tests and assign it to me?

[foolip]

@gsnedders, web-platform-tests/wpt#4822 is all yours :)