Clarify how to handle removed pointer capture element

[masayuki-nakano]

According to #526, when the override element is removed from the DOM, the pending element should be set to null and according to the other Pointer Event dispatching rules, browsers shouldn't dispatch lostpointercapture element on the removed target when the next processing of pending pointer capture.

However, what should happen if the node is moved to somewhere, especially when the node is moved into another document before processing the pointer capture. Should lostpointercapture be fired on the temporarily removed target? However, I feel it's odd that the event is fired on another document in the latter case.

[mustaqahmed]

In my interpretation, "moving an element" means removing the element first then appending it elsewhere, and this is true even for a move within the same document. The current spec text seems to cover this, no?

[smaug----]

move is in my mind only https://dom.spec.whatwg.org/#dom-parentnode-movebefore, these days at least :)

[masayuki-nakano]

Oh, I didn't know the API in @smaug---- 's comment, yeah, moving a node should be a pair of "remove" and "add".

Well, I was confused at the text in 9.5 Implicit release of pointer capture:

When the pending pointer capture target override is no longer connected [DOM], the pending pointer capture target override node SHOULD be cleared.

Okay, so, does this clearing happen without processing the pending pointer capture? I understood this as happening when processing the pending pointer capture.

[masayuki-nakano]

Ah, so, unclear scenario is:

1. A.setPointerCapture(n)
2. B.setPointerCapture(n)
3. lostpointercapture event listener moves B (then, pending element should be cleared)
4. Then, gotpointercapture should not be fired on B even if it's connected

Right?

Oh, but how do you handle this case?

1. A.setPointerCapture(n)
2. B.setPointerCapture(n)
3. lostpointercapture event listener moves B (then, pending element should be cleared) and call B.setPointerCapture(n) again.
4. Then, gotpointercapture should not be fired on B?

In theory, the pending pointer capture should be handled at next time and gotpointercapture should not be fired. However, it's hard to distinguish whether the B was removed temporarily or not...

[masayuki-nakano]

Oh, but how do you handle this case?

I think pending pointer capture element which is becoming new override element should be stored in wider scope and cleared if it's removed even temporarily.