[css-animation-worklet] The steps of registerAnimator mix types in a weird way

[bzbarsky]

https://drafts.css-houdini.org/css-animationworklet-1/#dom-animationworkletglobalscope-registeranimator step 5 does:

If SameValue(prototype, StatelessAnimator)

where prototype is the .prototype property of the constructor that was passed in, and StatelessAnimator is ... what exactly? It links to the interface definition. Does it mean the interface object? The interface prototype object? Something that has the interface prototype object on its prototype chain (i.e. checking that the passed-in thing is a subclass of StatelessAnimator)? Something else?

I suspect it means the subclass thing, but that's totally not what it says.

[majido]

You suspicion is correct. The intention was to verify "subclassing". I had trouble writing the spec text to do this correctly even after looking and other specs that meant to do similar thing. I have captured the difficulty in issue #850 and the fact that our current approach for using IsSameValue is incorrect.

In the most recent Houdini F2F heycam@ suggested a way we can spec this. But more importantly we resolved to drop these two superclasses anyways so we don't need to check the inheritance chain.

I have not made the changes to the specification yet but Blink implementation does match the most recent resolution (i.e., do not check the superclass). I will edit the spec based on most recent resolution which should also bring Blink implementation in line with the specification.

[bzbarsky]

Thank you, I'll take another look once the spec updates happen.

[majido]

I believe this is fixed by 48d9fe4

We no longer check the prototype of the stateless and stateful animator.