Pointer Events WG Voice Conference -- 10 Sep 2013

<scribe> ScribeNick: ArtB

<scribe> Scribe: Art

<smaug> coming

<rbyers> I'm on my way - having telco issues

Tweak agenda

AB: I posted a draft agenda on September 9 http://lists.w3.org/Archives/Public/public-pointer-events/2013JulSep/0022.html.
... Olli suggested in IRC we include bugs 22890 and 22891 today and that seems like a good addition. I propose we take them after Implementation status and Testing.
... Any objections?

[None]

AB: any other change requests for today's agenda?

CR implementation status

AB: the last time we discussed impl status was 30 July http://www.w3.org/2013/07/30-pointerevents-minutes.html#item02
... Since the draft agenda was published, we got a short update re Polymer from Daniel Freedman and good news from Jacob.
... Let's start with Jacob

JR: if folks want to test but don't have access, there will be update for Win 7 but no annouced date

… supports PE for mouse just like Win 8.1

AB: Matt, Olli, what's the latest on Gecko?

OP: touch action part we are waiting

… not yet done

RB: there was a thread in Bugzilla re touch actions

… is there consensus on Mozilla side?

OP: we are looking at it

RB: we found this is the hard part

… at least in Blink

OP: agree this is hard

RB: does touch action apply to touch event?

OP: we haven't discussed that

RB: we need to think about compat for these two

… I put a link to my design in Moz bugzilla bug

… I propose a new CSS property

… but it hasn't been implemented

… but that's in scope for Web Events WG

… we need to implement our proposal and test before bringing to standardization

<rbyers> regarding touch action in mozilla: in particular the issue is what the performance implications are - what blocks the main thread...

RB: re Blink, to get touch action work need a reliable touch system

… the hard part is the hit testing on the off thread

… we had an impl but it's busted

… I am now getting some more resources for touch action

… I think we have hit testing in a good place

… Now we need to rearchitect gestures and then start on touch action

<rbyers> yikes

RB: there was some discussion about adding YA property to Navigator

… for PE v2, think we want to think about a device query system

JR: you mean maxTouchPoints

RB: yes, that's right

JR: agree we need a longer-term way of handling this

<rbyers> maxTouchPoints approved for shipping in blink: https://groups.google.com/a/chromium.org/forum/#!msg/blink-dev/ayzxdztUlOQ/rd-z_Jo3ocIJ

DS: can we Agenda+ this

AB: ok with me

AV: what about polymer?

RB: Daniel posted some info

… we need to get status from jQuery

… we continue to use Polymfer in our projects

… a big question is how to handle IE6

… Scott and others said they would submit patches

… but they haven't done that yet

AV: when do you expect touch action to be done Rick?

RB: not sure; depends on "land"

… need to go thru reviewers, etc.

… at least a month away

Test Suite status

AB: yesterday Jacob committed some tests https://github.com/w3c/web-platform-tests/pull/324 mirrored to http://w3c-test.org/web-platform-tests/submissions/324/pointerevents/ and he updated the Assertion table http://www.w3.org/wiki/PointerEvents/TestAssertions
... Thanks Jacob!
... who can commit to reviewing Microsoft's tests?

… should we split them up?

DS: that makes sense

AB: about 20 or so files

JR: we will submit a few more files

MB: I can review some

OP: I can't commit now

CC: I can do some

RB: I can review some too

AB: and I'll take some

<scribe> ACTION: matt Divide up Msft's tests for review by Rick, Cathy, Art and Matt [recorded in http://www.w3.org/2013/09/10-pointerevents-minutes.html#action01]

<trackbot> Created ACTION-45 - Divide up msft's tests for review by rick, cathy, art and matt [on Matt Brubeck - due 2013-09-17].

AV: are there other PRs?

JR: is there an easy way to know the set of PRs for pointer events

<scribe> ACTION: barstow followup with Tobie re getting notifications for PE tests [recorded in http://www.w3.org/2013/09/10-pointerevents-minutes.html#action02]

<trackbot> Created ACTION-46 - Followup with tobie re getting notifications for pe tests [on Arthur Barstow - due 2013-09-17].

AV: I can review tests submitted by others

AB: ok
... Scott can now go through the TTWF tests with Dave http://lists.w3.org/Archives/Public/public-pointer-events/2013JulSep/0022.html aka Action-44 https://www.w3.org/2012/pointerevents/track/actions/44.
... after Scott has completed his action/analysis, we should have a reasonable idea of the coverage and holes

AV: there are 2 gaps

… assertions with no TAs

… features with no tests

AB: anything else on testing for today?

AV: re the TA wiki, some are marked as "X has written" but there is no link

<scribe> ACTION: barstow followup on the TA assertions to determine why there are some missing links to PRs/Submissions [recorded in http://www.w3.org/2013/09/10-pointerevents-minutes.html#action03]

<trackbot> Created ACTION-47 - Followup on the ta assertions to determine why there are some missing links to prs/submissions [on Arthur Barstow - due 2013-09-17].

Bug 22890 - It is not clear why navigator.pointerEnabled is needed

AB: 22890 https://www.w3.org/Bugs/Public/show_bug.cgi?id=22890 was submitted by Olli on August 6.

OP: I think the bug report is clear re the issue

RB: yeah, this comes back to what I said earlier re using Navigator

… I may not get approval to add this if window.PointerEvent can be used

… with v8, can't dynamically add properties

<sangwhan> Any particular usecase that makes the navigator member useful?

… why was pointerEnabled added?

JR: our original plan was to only support PE on Win 8+

… so this was created to determine if PE would fire on a particular platform

… later we added PE to Win 7

… so the reasoning is a bit moot at this point

… With our compat research, we have found pointerEnabled being used

… so if removed, would break some sites

… I agree with not putting stuff on Navigator but think it can be useful in the scenario I mentioned earlier

RB: Chrome's PE plan is to always support them

… perhaps we will need to disable in some cases

JR: on XP, follow the same pattern we use

<sangwhan> Chromecast or other TV/STBs comes to mind as one usecase that may not want to fire PE

RB: at some point Chrome will switch to use PE on Win8

… so for Blink, I will probably have a hard time selling Navigator.pointerEnabled

… probably need separate flags

… and see which sites break

… suspect it will be hard to add

OP: it will be hard to get added to Gecko

AB: it appears we don't have consensus on what to do

… do we leave it open?

RB: browsers could leave it out and only add it if really needed

… and we tell devs to use window.PointerEvent

JR: the timing now is problematic

… think it will be difficult to remove given some sites depend on it

… if the WG agrees to remove it, we could adjust our guidance

… but it will remain in our platform

RB: it's too bad we didn't catch this earlier

<sangwhan> Considering how fast library/framework devs react to spec changes I'm not sure if this is really going to be a problem, if there are open libraries that don't change reaching out doesn't take too much time..

OP: we should make sure documentation says to use window.PointerEvent (and not Navigator.pointerEnabled

RB: I don't think we will be able to add it until we can show/prove compat

AB: is there a test for this now?

<jrossi> I think Flipboard.com is an example that breaks without pointerEnabled

JR: not in our submission, perhaps TTWF submissions

AB: think this will be a problem re testing the CR

OP: should we add something to the spec re this "at risk"?

RB: we could advocate checking window.PointerEvent and then also check for pointerEnabled

JR: think we need make a call and then update the guidance

MB: do these sites already support the unprefixed version?

JR: yes, there are already some sites using pointerEnabled

<sangwhan> Do we have data on which sites?

MB: these sites using prefixed will need to change anyway

<rbyers> sangwhan: Jacob mentions flipboard.com and indeed I see that in their code

<scribe> ACTION: barstow add a link to bug 22890 that points to the 10-Sep-2013 discussion [recorded in http://www.w3.org/2013/09/10-pointerevents-minutes.html#action04]

<trackbot> Created ACTION-48 - Add a link to bug 22890 that points to the 10-sep-2013 discussion [on Arthur Barstow - due 2013-09-17].

Bug 22891 - Mechanism to differentiate pointer input from multiple users/devices missing

AB: 22891 https://www.w3.org/Bugs/Public/show_bug.cgi?id=22891 was submitted by Sangwhan on August 6.

<jrossi> http://lists.w3.org/Archives/Public/public-pointer-events/2013JanMar/0223.html\

<jrossi> http://lists.w3.org/Archives/Public/public-pointer-events/2013JanMar/0223.html

RB: we talked about this a couple of times

… this is the wii remote case

… we agreed we need a solution for this some day

… i.e. something we do in v2

JR: I just dropped in links to previous discussions

… think this is broader than just PE

… perhaps we need a new spec of diff spec like UI Events

<sangwhan> The root problem should probably be handled in the scope of UIEvents

AB: is there an action for sangwhan to move this bug to UI Events spec?

<sangwhan> No, but I can do it

<sangwhan> Give me a action, I'll contact Travis

AB: any objections for that resolution? i.e. Sanwhan move 22891 to UI Events?

<rbyers> I'd like to include scenarios like "is there a physical keyboard attached" - I think it's the same sort of 'input device query' API...

[ None ]

<scribe> ACTION: moon move 22891 to UI Events [recorded in http://www.w3.org/2013/09/10-pointerevents-minutes.html#action05]

<trackbot> Created ACTION-49 - Move 22891 to ui events [on Sangwhan Moon - due 2013-09-17].

<smaug> (queries are somewhat privacy sensitive)

RB: I think we have a related entry in our v2 list

<asir> here http://www.w3.org/wiki/PointerEvents/UseCasesAndRequirements

RB: yeah, that's it

<scribe> ACTION: barstow add a link to the v2 doc to the PE main page [recorded in http://www.w3.org/2013/09/10-pointerevents-minutes.html#action06]

<trackbot> Created ACTION-50 - Add a link to the v2 doc to the pe main page [on Arthur Barstow - due 2013-09-17].

AB: Sangwhan, when you move 22891 to UI Events, please include a link to the v2 UC doc

<sangwhan> Art, will do.

detecting browser capabilities

DS: so this is not a new problem

… we anticipated this early on (DOM specs)

… the solution at the time was "hasFeature"

… I understand it got misused

… and there were big probs

… one couldn't count on using it (reliably)

… if there was even a "little bit" of support for a feature, it would return "true"

… lots of impls said yes when there was no support at all

… For D3E, we worked out a proposal to use strings

… that can be namespaced and based on support for features

… F.ex. to check for attr X, could use dot notation and check True/False

… Is there any way this can really be done in a reliable and compatable way?

RB: agree we need a general mechanism for feature detection

… for the most part, I think what we have today works

… can be problems with events

… Not clear we want to add a bunch of complexity

… there will always be a way to check if an object exists or not

… not convinced the benefit of adding a second system is worth the cost

DS: there are problems with just checking an object

RB: well in Bink, we don't object an object unless it is complete

DS: browsers need to be more strict

… Some things are tricky to test for

… f.ex. modenizer creates objects just to see a feature exists

… some features are hard to detect

… Has "hasFeature" been deprecated at this point?

JR: not sure if hasFeature is being used for pointer events

RB: we are debating if we need an additional mechanism for detecting PE or not

… if so, do we use Navigator, do we use hasFeature, etc.

… I hope we can just use window.PointerEvents

AB: is there a conclusion or followup for someone?

DS: don't think so

RB: think we still are at the question about is pointerEnabled needed or not

DS: the original design was each spec would define the string for their feature(s)

… but I think we need to decide pointerEnable or not

… and then if we need it, consider some more general solution

AoB

DS: W3C has changed its policy re normative references

… we have a more pragmatic approach now

… rather than looking at a spec in totality, it is now possible to view the references in parts

… A consequence is this means Web Events can move to REC

AV: we are wondering about a f2f meeting

… f.ex. to review tests, add tests

… what do you think

… can people think about that

RB: I am a fan of f2f meeting but we need to think about the timing

… might make more sense to meet after we have more than one impl avail

AB: those are good points
... It will be difficult for me to meet before TPAC

DS: same for me re logistics

… agree f2f meetings for thinks Asir mentioned makes sense

MB: October is hard for us too

RB: if we have `done` impls, would it make sense to get together then?

MB: yes, I think so

… but now we have unlanded patches

AV: well the 8 week notice does cause a problem

DS: anyone going to TPAC?

AB: I plan to go

<jrossi> I'll be at HTML5DevConf :-)

RB: there could be a conference when we could co-locate

<sangwhan> I don't know yet

DS: HTML5DevConf could be a rallying point

… even if informal

JR: yes, I could meet in that timeframe, even if informal

RB: we could demo the polymer pollyfill

… but that week won't work for me

… Blink conf is another possibility

AB: when is Blink conf?

RB: Sept 24-25

AV: so my summary is that we need more than 8 weeks

… would be good to try to co-locate with some conf

DS: if we do meet, I would like to have an open meeting for people outside the meeting

<rbyers> jrossi: nope, it's this: https://groups.google.com/a/chromium.org/forum/#!searchin/blink-dev/blinkon/blink-dev/vKAVn47Cn-k/4mPyBAXsrKsJ

<rbyers> pretty small scale

AB: so next meeting will be when we have sufficient topics

… Meeting adjourned

- DRAFT -

Pointer Events WG Voice Conference

10 Sep 2013

Attendees

Contents