Minutes from PEWG meeting 9 October 2024

Dear all,

the minutes from today's meeting are at 
https://www.w3.org/2024/10/09-pointerevents-minutes.html and copied below:


PEWG
09 October 2024

Agenda: 
https://www.w3.org/events/meetings/66591f6b-6694-4f90-b23d-bf8f1b9dda8a/20241009T110000/
IRC log: https://www.w3.org/2024/10/09-pointerevents-irc

Attendees
Present: adettenb, flackr, mustaq, Olga, Patrick_H_Lauke, smaug

Chair: Patrick H. Lauke
Scribe: Patrick_H_Lauke



* Limit the precision of floating point event fields w3c/pointerevents#517
* Ensure predicted events only use input from the current partition 
w3c/pointerevents#518
* [PointerEvent algorithms] Order of boundary events w3c/pointerevents#519
* [touch actions] handwriting manipulation type to distinguish panning 
w3c/pointerevents#516
* Coalesced and predicted event attributes within an untrusted event 
w3c/pointerevents#514
* Ambiguity of the value of the button property for the click event 
w3c/pointerevents#513
* Triage unlabelled issues https://github.com/w3c/pointerevents/issues
* Meta-issue: update WPT to cover Pointer Events Level 3 #445 
w3c/pointerevents#445




[quick round of introductions for the benefit of Olga and Adam, new WG 
members joining us from Microsoft]



# Limit the precision of floating point event fields w3c/pointerevents#517
Rob: meant to write something up, but as mentioned in previous meeting: 
location should be at least pixel level; and we need to be careful about 
potential banding if we artificially limited precision that would then 
lead to jumps (e.g. pressure)

ACTION: continue iterating, Rob to write up a few thoughts on the issue



# Ensure predicted events only use input from the current partition 
w3c/pointerevents#518

Patrick: i promised that i'd monkey-patch this to at least make it clear 
what we mean by "past" (i.e. the preceding points of the current 
gesture/interaction, not "when the user last visited this a few days 
ago"). suggest i'll do this for next meeting, then we have something to 
expand further

Rob: also limit it specifically to "current webpage" or similar, be very 
specific, to make it clear devs won't have more access to things than 
they should

ACTION: Patrick to make first pass PR to clarify



# [PointerEvent algorithms] Order of boundary events w3c/pointerevents#519
mustaq: UI events spec wants to become more algorithmic, and that showed 
up a few discrepancies with our own handling. this was before TPAC. but 
at TPAC it was decided that UI events will handle mouse events (?)

<mustaq> https://www.irccloud.com/pastebin/wmpUV7lT/

https://www.w3.org/2024/09/WebAppsWG-TPAC2024-Meetings-Minutes.pdf

mustaq: there are four of these algorithm issues altogether. they're 
more longer-term. not for Level 3

Olli: agree this is more for Level 4, Next ... post Level 3

Rob: we have clear direction, we just need to make sure it's all handled 
consistently

Adam: is it that mouse events currently are getting turned into pointer 
events, for enter/leave...

mustaq: not a problem of event per se, but which spec specifies the behavior

Rob: developers won't see any difference. it's just which order, which 
timing, events are being fired



# [touch actions] handwriting manipulation type to distinguish panning 
w3c/pointerevents#516
Adam: in short, we need a way to separate handwriting from scrolling 
behavior

Adam: e.g. an app adds a toggle to explicitly switch between handwriting 
and regular pointing device for their stylus

Adam: one proposal was to do this as an attribute in HTML. another idea 
was to piggy-back on touch-action as a new value to differentiate 
between handwriting and panning

Adam: no way currently to allow granular control

<mustaq> w3c/pointerevents#203

Rob: the way we built stylus on android is that handwriting is 
equivalent to scrolling (?)

Rob: not sure if 203 is same issue. when you put stylus down, you CAN 
pan, you CAN write, nothing for author to stop one or the other

Adam: you can only limit pan to one direction, but not completely 
suppress pan AND allow handwriting

Rob: pan and manipulation might disable handwriting

Rob: problem is that touch action is applied as a bitmask down the tree 
... if the parent limits to just panning, for instance, it would then 
disable handwriting further down

Olga: so we won't be able to enable handwriting on a child if parent has 
limited to just pan

Rob: if we don't include handwriting as a concept as part of pan

Olga: as Adam pointed out, this might be more a problem for older pages, 
but not for new ones

Adam: yes, new ones may already take that into account. so adding a new 
value for handwriting to touch-action sounds like the way to go

Rob: might be worth also thinking about other actions like text selection

Adam: should it be included in the manipulation set?

mustaq: in Chrome manipulation is pan and zoom

Rob: i think it should probably be in manipulation

Rob: we should have some other property to determine whether you want 
... how you want to treat these devices, but maybe touch-action is 
sufficient. in other issue (512) i talk about how we might want to allow 
mice to pan. that's more a "how to treat this device"

Olga: in future it could allow handwriting with mouse (in ref to 512)

mustaq: in reference to 203, that was a wider idea of moving away from 
touch-action to a few more specific properties

Olli: is there a situation where you'd also want different 
pointers/styli to behave differently?

Adam: some styluses support things like a toggle/switch to trigger 
different behaviors, stylus with eraser. not sure what that would look like

Olga: to be clear, we just want to allow author to specify that they 
want to allow handwriting with a pointer (stylus, finger) rather than 
interpreting it as a pan (?)

mustaq: what to do when a device doesn't support handwriting? should it 
fall back to then allowing pan?

mustaq: what i imagine is a page with lots of inputs, and you want to 
allow an author to say "for touch, just pan; for pen, make it do 
handwriting"

Adam: would be good to allow both, but have handwriting take precedence. 
Would be useful to allow defining different behaviors for different 
pointer types too

<mustaq> w3c/pointerevents#203 (comment)

mustaq: there's no consensus there, but the issue there mentions this 
idea of making it more granular / per pointer type

<mustaq> pointer-action: touch(pan-y), pen(none);

ACTION: iterate further on the issue w3c/pointerevents#516 - Adam to 
flesh out further thoughts/ideas (also in light of 203)



# Coalesced and predicted event attributes within an untrusted event 
w3c/pointerevents#514

mustaq: my concern was that authors may trust the trusted bit too much, 
even though the list comes from an untrusted event...

ACTION: mustaq to propose a PR with clarification, Olli to review



# Ambiguity of the value of the button property for the click event 
w3c/pointerevents#513

Olli: ...in this case click needs to follow UI events

mustaq: the special case i proposed will be breaking...

Olli: we can't change behaviour, just need to clarify it

ACTION: Patrick to attempt first PR to clarify/document this



# Triage unlabelled issues https://github.com/w3c/pointerevents/issues



# Meta-issue: update WPT to cover Pointer Events Level 3 #445 
w3c/pointerevents#445

mustaq: think we only have one left at this point. Rob landed it, but it 
was reverted.

w3c/pointerevents#300

ACTION: Rob to review his WPT (and why it might have been reverted)


-- 
Patrick H. Lauke

* https://www.splintered.co.uk/
* https://github.com/patrickhlauke
* https://flickr.com/photos/redux/
* https://mastodon.social/@patrick_h_lauke

Received on Wednesday, 9 October 2024 16:09:19 UTC