Minutes from PEWG meeting 25 October 2023

Dear all,

the minutes from today's PEWG meeting are available at 
https://www.w3.org/2023/10/25-pointerevents-minutes.html and copied below:


PEWG
25 October 2023
Agenda: 
https://www.w3.org/events/meetings/6246bc85-4dae-43a8-a50c-9bc5a0829585/20231025T110000/
IRC log: https://www.w3.org/2023/10/25-pointerevents-irc

Attendees
flackr, mustaq, Patrick_H_Lauke, plh, smaug

Chair: Patrick H. Lauke
Scribe: Patrick H. Lauke, Patrick_H_Lauke



* wide review
* Rob as co-editor
* wide review and charter
* Review outstanding v3-blocker issues 
https://github.com/w3c/pointerevents/issues?q=is%3Aissue+is%3Aopen+label%3Av3-blocking
* interop 2024 effort
* sideline pointervent id


# wide review

Patrick: w3c/pointerevents#482

Patrick: been slack after TPAC with vacation and another conference, but 
am now looking at this

Patrick: extra wrinkle is our charter runs out, so coordinating with PLH 
about extending it again

Patrick: want to get this done on my end ASAP



# Rob as co-editor

Patrick: you may have seen, but now Rob officially co-editor as well

w3c/pointerevents#488



# wide review and charter

plh: timeline in charter is outdated, so don't know what to put

Patrick: if I kick off wide review now, how long will it take realistically?

plh: usually 2 months, maybe even 1

Patrick: safer to say 2 months

plh: when do you expect wide review to be sent out?

Patrick: want to get it done tonight, tomorrow, this week at least

plh: so that puts us into January. i'll update timeline and charter



# Review outstanding v3-blocker issues 
https://github.com/w3c/pointerevents/issues?q=is%3Aissue+is%3Aopen+label%3Av3-blocking

Patrick: where are we up to with #477 w3c/pointerevents#477 ?

mustaq: we have comment that we need one implementer. chrome will 
hopefully have this. also we need a WPT for shadow DOM

Olli: yes we need to define how that is supposed to work

mustaq: anyone working on that WPT?

Olli: i'm not on that one, no

<mustaq> I am not either!

Olli: hoping that while writing the test, it will become clear what the 
right approach should be

Olli: it should be event path though, as mentioned

Rob: yes, it's mentioned in various places, like UI Events, but they 
don't mention shadow DOM

Olli: was thinking DOM spec, if there's a nice algorithm...

Rob: we need to define event path somewhere, and then refer to it. then 
we can just refer to parent in event path

Rob: the modification ... you need to do this not while processing the 
event, happens outside of the event processing. you could refer to the 
historical event, but realistically want to follow same logic as an 
event that would have targeted that element

mustaq: the event path is constructed during the dispatch, so will be 
tricky...

Rob: we should use same rules as the constructed event path for the 
thing that you're now considered to be over

Rob: i see you (Olli) have added a link to the DOM spec event path

Olli: yeah but sorry if we can't really use any of the algo. because 
it's to dispatch an event to a target. then there's the composed (?) path

Patrick: so we still iterating...

Rob: looks like the event dispatching constructs the path, so maybe can 
extract that

<smaug> https://dom.spec.whatwg.org/#dispatching-events

<flackr> https://dom.spec.whatwg.org/#dispatching-events

Patrick: are we defining stuff that we shouldn't?

Rob: I want to take 5.9 and just extract/reuse that

Rob: you keep stepping up until you get to the root, and that's what 
defines the path

mustaq: so we basically say ... "conform to 5.9" so we don't need to 
change the DOM spec

Rob: that would be the correct test, but we probably need to say 
something more direct in our spec. when target is removed, we follow 
this algo until we find a still attached node

Patrick: just wondering, as this is the last substantive change, should 
we hold off with wide review until we have this?

Olli: yeah, might be true

<mustaq> w3c/pointerevents#487

Patrick: we can probably go for review before there's implementation and 
WPT, but we probably should make the change to spec first before asking 
for review

mustaq: filed the above bug which is related...

Rob: I thought our spec did say what to do

Olli: yeah so did i

mustaq: feel free to comment in issue, maybe i missed something

Rob: section 9.5 implicit release of pointer capture says that when 
target override is no longer connected, we clear ...

Olli: need to look whether it's talking about connected, disconnected, 
or removed

mustaq: ... we're expected to fire lostpointercapture...

Rob: I believe so

Olli: this is the behaviour that browsers had

Rob: I thought so yes

mustaq: the timing is not clear. the get or lostpointercapture is fired 
in a lazy manner

mustaq: should we fire immediately after DOM change or should it wait?

Rob: it also talks about the pending pointer capture in 9.5. my reading 
is that it should be immediate

mustaq: i think second para mostly addresses that, but timing could do 
with clarification

Rob: we don't want to fire immediately as that makes it sync, but we 
should treat it as lost pointer capture as normal and queue it

Olli: do we queue a task or run as soon as possible

mustaq: lazy mechanism might not see it as a change that needs dispatching

Rob: when does the pending stuff get sent

Rob: ... do we expect a notification when we don't get pointercapture?

mustaq: lazy mechanism mentions something...

<mustaq> 
https://w3c.github.io/pointerevents/#process-pending-pointer-capture

mustaq: i think this section (link) mentions the lazy mechanism

Rob: right now, i f you set pointercapture, if you remove node before 
the next event, there will be no notification that you didn't succeed in 
getting pointercapture

mustaq: which is fine right?

Rob: not great from a developer perspective

mustaq: it's not a got or lost, it's a cancel pending request

Rob: you could send lostpointercapture, but then it wouldn't match up to 
a gotpointercapture

Rob: maybe it's something we add in future...

mustaq: back to timing, do we agree...

Rob: it says "otherwise clear the pointercapture target override". is 
that when we fire lostpointecapture.... oh no it's step 1

Rob: if the node was removed, it would be in step 1 that we detect that 
the capture target is null

Rob: that's the point where we would fire lostpointercapture

mustaq: so there we should immediately set pending to null

mustaq: please comment in issue, then we can resolve this easily

ACTION: review #487 and hopefully close if already covered

Patrick: getting back to #477 who would like to take a stab at this

PAtrick: would be keen to get the small change in our spec

Rob: not sure how easy it is to get change into DOM spec...

Patrick: but just the part for our spec...

Rob: I can try and just reference...

ACTION: Rob to review/make necessary change to PE spec relating to #477



# interop 2024 effort

mustaq: want to highlight proposal for pointervent/mouseevent

<mustaq> Interop 2024 carryover proposal for Pointer Events and Mouse 
Events:

<mustaq> web-platform-tests/interop#472

<mustaq> It would be great to see other browsers adding to the list.

Patrick: so is this for Olli to be aware of and have a look at? Yeah, cool

ACTION: Olli to review interop 2024 carryover proposal for PE and Mouse 
Events



# sideline pointervent id

Patrick: 
https://groups.google.com/a/chromium.org/g/blink-dev/c/3eU-AHH8x4k/m/rCyxcYnQAgAJ 
intent to ship in chrome

Patrick: assume this is stuff for future versions of PE?

Rob: yeah that seems reasonable

Patrick: not trying to put any of you on the spot, just wondered if you 
had more inside knowledge

Patrick: assume this is what came out of the presentation we had a while 
ago from... Microsoft? Wacom? ... about their idea to store id more 
permanently to then open up possibility of storing preferences

Rob: looks like that on that intent thread two of the owners pointed out 
it should have a spec

Patrick: cool, just no action right now, just something to be aware of 
that it might come in for future versions

Rob: FWIW i reviewed early work on this, and made sure things were 
designed in similar way to what we already have. but yes we'll evaluate 
it properly when it comes to us




Summary of action items
* review #487 and hopefully close if already covered
* Rob to review/make necessary change to PE spec relating to #477
* Olli to review interop 2024 carryover proposal for PE and Mouse Events


-- 
Patrick H. Lauke

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

Received on Wednesday, 25 October 2023 15:53:36 UTC