Minutes from Pointer Events WG call 2 October 2019

Dear all,

minutes from today's call available on 
https://www.w3.org/2019/10/02-pointerevents-minutes.html and posted below


PEWG
02 Oct 2019
Agenda

Attendees
Present
patrick_h_lauke, plh, NavidZ, Olli_Pettay
Regrets
Chair
Patrick H. Lauke
Scribe
patrick_h_lauke
Contents
Topics
any relevant corridor/side discussions at TPAC (since we didn't have an 
official meeting there this time)
merging extension into main spec 
https://w3c.github.io/pointerevents/extension.html
tiltX and tiltY aren't as helpful as just having altitude angle 
https://github.com/w3c/pointerevents/issues/274
Setting pointer capture on an element in a parent frame when the pointer 
was made active in a child frame 
https://github.com/w3c/pointerevents/issues/291
Changing the DOM hierarchy while handling a "pointerenter" event 
produces significantly different results across browsers 
https://github.com/w3c/pointerevents/issues/285
Summary of Action Items
Summary of Resolutions
<scribe> Scribe: patrick_h_lauke

<smaug> webex is so not working for me

we'll wait for you smaug

any relevant corridor/side discussions at TPAC (since we didn't have an 
official meeting there this time)
Navid: proposal that microsoft had to send click/auxclick etc as pointer 
events

chrome has this behind experimental flag

for a few months

not had regression/bug reports, either because they're fine or low usage

we actually went with more aggressive version, where click etc are 
actual PE with fractional coordinates

we should define what it means for the other properties in PE

e.g. pointerType, if it comes from the keyboard

we could set to empty string

need to come up with a number for pointerId

wonder if we should say pointerId '0' value is the "we don't know" 
origin/value (e.g. generated by keyboard)

wondering what others think about that

we have normative language about pointerId can be recycled/reused. what 
happens then with click that comes after pointerup/pointerdown/etc

define WHEN ids can be recycled (until the whole event sequence is 
dispatched, including compat mouse events and click)

Olli: right now pointerId zero can be used for anything else so can't 
really use it

gecko uses zero already so can't be made special just now

wonder if we should use max value (it's LONG not Unsigned). could we use 
"-1" ?

Navid: we should double-check with other implementations if they ever 
assign negative pointerId

Olli: sounds like a good initial idea [pointerType empty and pointerId=-1]

Patrick: also like conceptually that pointerId="-1" kind of implies 
"this is (likely) not an actual pointer" as it feels conceptually a bit 
dirty that keyboard would fire a pointer event

Navid: ok, we'll run a test in chrome, do a PR to spec, and we look if 
we get regressions etc
... other topic discussed was longer discussion from Daniel Libby about 
pointer event trail - he filed issue about it in GH

asking underlying API to generate trail under pointer

[gives more details about the proposal - web app declaratively 
requesting platform draw a particular trail with certain characteristics]

reduced latency of this approach

saw demo on iPad at 120Hz

and on MS tablets. just drawing "shadows"

half of latency on Surface (at 50Hz) compared to iPad (120Hz)

but if only on windows, not the way to go

don't recall if Apple said they have similar API

Google plan to talk to Android etc platform to check if such a low-level 
API that can be hooked into exists

question is heuristics, legality, and whether it fits into the PE domain 
or not

Olli: it probably fits into PE as you should be able to use it with all 
pointers...but if not all OSs support, could/should browsers support it 
themselves (sending coords to their internal compositor, improving 
latency/performance a bit)

is that doable in chrome

Navid: forgot to mention that, already discussed with team, and it would 
also work/improve performance

worth shipping, EVEN if it was only platform native on Win and other 
platforms do it in browser?

Olli: still need to iron out issues like cross-origin iframes etc

Patrick: wondering if it's not scope creep to cram this into Pointer 
Events spec itself - thinking for instance PointerLock spec which is 
separate

Navid: concern also with exclusions/IP, so it might prevent us from 
doing it as part of PE

but it should build on top of PE

<plh> --> 
https://www.w3.org/2018/12/pointerevents-charter.html#out-of-scope Out 
of Scope

Patrick: agree, might be worth making extension/incubating

further updates from TPAC?

[Olli mentions pen action idea that was discussed]

Navid: event pointertype CSS definition (?) expanding touch-action to 
have support per pointer type. will file something on mailing list/gh

Microsoft will write an explainer for it

Mustaq told me that CSS folks wouldn't like this to be part of CSS, as 
these are not about rendering but targetting of events

merging extension into main spec 
https://w3c.github.io/pointerevents/extension.html
Patrick: are we happy to go ahead with merge, and then go for First 
Public Working Draft stage

Navid: we have some aspects that are handwavy in current text in 
extension. are we happy with it for now, or do we want to make it more 
algorithmic like similar specs, rewrite portions to formally and 
precisely do it

Olli: that would all depend on Gary's UI event rewrite

Navid: when PE started, it was right before UI events stuff. we started 
having UI events relying on us in terms of dispatch algorithm

one thing we can do with FPWD is be ok with what we have now and hope 
for best/tweak

plh: do we have any test for this?

Navid: yes

plh: then i say merge it

Olli: +1

Navid: going to send PR and have Webkit folks also look at it. deadline 
of about 1 month?

plh: yes and then we can publish FPWD
... when you do merge, can you also move test as appropriate?

Navid: yes, will sync

tiltX and tiltY aren't as helpful as just having altitude angle 
https://github.com/w3c/pointerevents/issues/274
PAtrick: this is mostly a wrist-slap for myself as i promised i'd do a 
basic PR, but not got around to it. hope to have for next meeting

Setting pointer capture on an element in a parent frame when the pointer 
was made active in a child frame 
https://github.com/w3c/pointerevents/issues/291
<NavidZ_> https://github.com/w3c/pointerevents/pull/300

Navid: i had this pull request...

that's current proposal, it seems that people are ok with it?

Olli: commented on it but not reviewed again

Changing the DOM hierarchy while handling a "pointerenter" event 
produces significantly different results across browsers 
https://github.com/w3c/pointerevents/issues/285
Navid: briefly discussed with Gary

will be dealt with in UI events

bigger undertaking than just PE

suggestion: don't touch it for now

Olli: was also mentioning to Gary, and it's UI events issue about better 
algorithmic spec

do we have a matching UI-events issue? if so we could just link

Navid: don't see an issue, want to create one?

Olli: i'll create one

Patrick: and we leave ours open to see what comes back


-- 
Patrick H. Lauke

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

Received on Wednesday, 2 October 2019 15:52:12 UTC