messageerror event should use ExtendableMessageEvent

[bakulf]

Spec says:

If this throws an exception, catch it, fire an event named messageerror at destination, using MessageEvent, with the origin attribute initialized to origin and the source attribute initialized to source, and then abort these steps.

but MessageEvent.source cannot contain the source if it is a Client. We should probably use ExtendableMessageEvent instead.

[jakearchibald]

Agreed!

[mkruisselbrink]

For the first place this sentence occurs (where the actual message uses ExtendableMessageEvent) definitely. For Client.postMessage using MessageEvent is correct (and consistent with the rest of the algorithm) of course.

[domenic]

Is the event really intended to be extendable?

If not, perhaps adding Client to the enum at https://html.spec.whatwg.org/multipage/comms.html#the-messageevent-interfaces would be a better choice.

[jungkees]

I think it'd make sense to allow lifetime extensions to messageerror events. Otherwise, we can't guarantee that the event handlers will run to completion.