Bug 203232 – Touch events are dispatched to document element when the original target is detached and preventDefault was called for previous event

• Component: UI Events
• Status: NEW

[Xidorn Quan]

Created attachment 381505 [details]
testcase

Please see the attached testcase. In the testcase, there is a box keep detaching and attaching itself every two seconds. And there is an event handler on `touchmove` on document which logs a "D", and a handler on the box which logs "T".

Steps to reproduce:
1. when the box is there, press on the box and move the finger around
2. wait until the box disappears, and move the finger around

Expected result:
You should see TTTT being logged because the touch event should goes to where it starts (which is the box).

Actual result:
You would see DDD being logged because the touch event gets to the document element instead.


This behavior doesn't match the spec, and any other implementation.

Also this only happens when there is `e.preventDefault()`. If you remove that line, then it correctly logs TTTT indicating the events go to the box.

[Xidorn Quan]

Also this is something new in iOS 13, and it doesn't happen on iOS 12.

[Radar WebKit Bug Importer]

<rdar://problem/56557577>

[Antoine Quint]

As far as I know, our implementation of touch events isn't meant to match any particular specification. You say there's a behavior change between iOS 12 and 13 and that's worth investigating.

We now implement Pointer Events. Do you see similar behavior with Pointer Events?

[Xidorn Quan]

> As far as I know, our implementation of touch events isn't meant to match any particular specification. You say there's a behavior change between iOS 12 and 13 and that's worth investigating.

I think the touch events behavior is covered by Touch Events spec https://www.w3.org/TR/touch-events/#the-touchmove-event which stats:
>  The target of this event must be the same Element on which the touch point started when it was first placed on the surface, ...

which matches all other browsers as well as iOS 12.

> We now implement Pointer Events. Do you see similar behavior with Pointer Events?

I haven't tried yet, and I'm not quite familiar with Pointer Events, but will have a look.

Actually I was a bit surprised that you decided to implement Pointer Events, as I thought WebKit was publicly against this.