Fix "relevant mutations" tests for <img> crossorigin attribute

[spinda]

These test cases are supposed to verify that changing an <img> tag's crossorigin attribute state causes the image resource to be reloaded (as per "4.8.4.3.2 Reacting to DOM mutations" in the HTML spec). However, they were set to expect a 'timeout', which doesn't make any sense: the expected result should be a 'load'. They were passing regardless of implementation correctness because the corresponding <img> tags were missing src attributes, which prevented load events from being generated.

See also: whatwg/html#4533

[zcorpan]

These are intentionally omitting src. They are testing that load and error are not fired as a result of mutating crossorigin when there's no src (or srcset/picture).

It could be useful to also have tests for mutating crossorigin with src, and certainly there could be source comments explaining things.

A previous PR that touched these tests: #2136

[spinda]

I see, that clears things up a bit. Thanks!

It could be useful to also have tests for mutating crossorigin with src

It certainly would, since both Firefox and Chromium fail these tests :)

I'll put the old tests back in addition to these "new" tests.

edit: I'm not sure why adding this comment caused the bot to ping a whole bunch of people all at once. Excuse me if I've done something wrong!

[zcorpan]

Thanks! The relevant mutations says "The element's crossorigin attribute's state is changed.", and these tests all change the state. (empty to "anonymous", or "foo" to "bar", would not change state.)

Can you report bugs on browsers that fail the new tests?

[zcorpan]

@wpt-pr-bot should have done its thing when the PR was created. I'm not sure why it didn't, but it has happened before.

[spinda]

Thanks! The relevant mutations says "The element's crossorigin attribute's state is changed.", and these tests all change the state. (empty to "anonymous", or "foo" to "bar", would not change state.)

Yep! The existing "crossorigin state not changed" cases cover the other side of this.

Can you report bugs on browsers that fail the new tests?

Can do!

[spinda]

Hmm, looking at the diff, seems that last push got mangled somewhere along the way. One sec...

[spinda]

Okay, fixed.

[zcorpan]

Thank you! Please add a comment linking to bugs for this.