Clarify that coalesced events list contain also a clone of the parent event

[smaug----]

Since the list below talks about "parent" event, perhaps just clarifying the sentence is actually enough here..

This PR is for #409

---

Preview | Diff

[flackr]

Minor nit: since the parent event can include slight interpolation / prediction it might be better to say something like including the event represented by the parent event?

[flackr]

Nevermind, it looks like the parent event should have the same coordinates

[NavidZ]

Yeah. I don't think the new statement is fully correct. The parent event might have different movementX/Y as @smaug---- also mentioned but it could also be resampled to what the user agent believes to be best representative of the current state (e.g. location, pressure and other similar stuff) of the pointer as @flackr mentioned. On the other hand the coalesced events are meant to be raw events received from the platfrom with no modifications on them.

[smaug----]

The current idea is to explicitly define parent event's movementX/Y to be relative to the previous parent event and the
movementX/Y of coalesced events to be relative to the previous event in a coalesced event list.

[smaug----]

I think one possible way to clarify this is to just say that if the coalesced events list was otherwise empty for trusted pointermove/pointerrawupdate, then it contains a clone of the parent event.

And then have totally separate definition how movementX/Y is computed for the events in the coalesced events list and for the parent events.

[patrickhlauke]

so, naive idea, can the wording just say something along the lines of "the last event in the coalesced events list corresponds to the "parent" event (with the same values for the various PointerEvent properties). This means that for trusted pointer events, coalesced events list will always have at least one event." or something along those lines?

and yes, then adding an informative note that other properties, such as movementX/movementY, of that last event are likely to differ from the parent event. or is that too handwavy?

[smaug----]

this is explicitly not talking about movement, since that seem to be separate issue altogether. Currently per specifications all the pointerevents should have movementX/Y 0 (whether or not the are in the coalesced events list). But the idea here is that this could work once movement is clarified elsewhere.

[patrickhlauke]

naive idea (as usual from me): how about defining that if the list does not have any events (i.e. there were no extra events that were coalesced into a parent), then the list will contain a clone of the parent event, and otherwise leaving it as is? would that cover the scenario where a developer wants to only use the coalesced events list?

edit: conceptually, I could even rationalise this in terms of the coalesced list containing all events that led the browser to fire the "proper" parent event. in the case where no coalescing happened, it can kind of make sense that it contains just the pure event that is also the parent event?

[patrickhlauke]

From call just now, idea from @flackr Right after this line, something along the lines of: the parent event is the result of all these events, with a hint that UA may resample or otherwise average/modify

[patrickhlauke]

parent event is a representation of the events in the list, including resampling etc, to the point where the UA thinks the parent is the most representative of the raw coalesced event stream's position etc

[patrickhlauke]

This is superseded by #440