Clarify `pointercancel` event target for drag start if the hit-testing result is different document from `dragstart` target

[masayuki-nakano]

The pointercancel event target is defined as same as the other pointer events by "fire a pointer event".

• If the pointer capture target override has been set for the pointer, set the target to pointer capture target override object.
• Otherwise, set the target to the object returned by normal hit test mechanisms (out of scope for this specification).

If pointercancel is caused by staring a drag session of mouse, it follows dragstart and usually it's not captured by the pointer capture API. Therefore, the target should be considered by the second one of the definition, i.e., considered with a hit-testing.

In edge cases, another document may be under another document which is different from the document received the preceding pointerdown and dragstart. In this case, pointercancel event should be fired on:

• root element or the document node which received pointerdown if the pointer is canceled outside the document
• the <iframe> element if the pointer is over a child document

The reason is, pointercanel is required for aborting to handle the pointer maybe started by pointerdown event listener.

[mustaqahmed]

Good catch, thanks!!

Let me try to dump my understanding of the general problem here:

• the pointercancel event could have been caused by "something" outside the document that received the pointerdown, and in that case the second bullet at "normal hit test mechanism" fails to define the target; and
• that "something" could be more than just a drag-start (e.g. it can be a pinch-zoom gesture where the second finger went down in a different document).

Was I able to capture your point, @masayuki-nakano?

[masayuki-nakano]

@mustaqahmed Exactly. On the other hand, @smaug---- found that the target of pointercapture for a drag-start is defined by the HTML spec. This should be referred by Pointer Events or be duplicated in the Pointer Events.