Pointer Events WG Voice Conference -- 26 Feb 2013

<scribe> ScribeNick: ArtB

<scribe> Scribe: Art

<smaug> Zakim: who is noisy

Getting started

AB: I posted a draft agenda yesterday http://lists.w3.org/Archives/Public/public-pointer-events/2013JanMar/0153.html. Any change requests?

JR: I enjoyed talking about PE last week @ W3Conf

… we can talk about that during AoB

AB: ok, we'll add that to AoB

Spec feedback by Alex Russell

AB: Alex submitted 7-8 bullets in his comments http://lists.w3.org/Archives/Public/public-pointer-events/2013JanMar/0110.html and they were submitted before the LC was published.
... we agreed during our February 12 call Alex's comments would be considered as LC comments http://www.w3.org/2013/02/12-pointerevents-minutes.html#item02.
... We could do a deep dive on some set of Alex's points or let the Editors reply first. I note one of Alex's comments is about the issue Rick raised about the semantics of pointerID which is on the agenda.

JR: I have a draft about half done re Alex's comments

AB: sounds good

Tweaking wording in Introduction by Rick Byers

<slightlyoff> apologies for not seeing this earlier. Wasn't aware there was a meeting

AB: Rick had some comments re the Introduction http://lists.w3.org/Archives/Public/public-pointer-events/2013JanMar/0127.html
... the LC already addresses at least some of Rick's comments but it appears there is also a request to embellish some set of the existing examples and/or add new example(s).

JR: 2 things

… I made most of the changes in the LC

… there is a bug I introduced that needs to be fixed

… and I'll do that

… Not sure what you mean by examples

AB: ok, I'll re-read that thread and reply accordingly

pointerType extensibility by Rick Byers

AB: On Feb 19, Rick started a new thread about pointerType extensibility http://lists.w3.org/Archives/Public/public-pointer-events/2013JanMar/0134.html.
... is this a request to change the API defined in LC for v1 or is this more a question about what we might want to do in v2?

JR: I talked to Rick about this last week

… we talked about diff ideas

… they all have + and -

… My conclusion is that there are 2 scenarios

… one is providing a better path for new devices

… we want them to have some compat with PEs

<slightlyoff> just joined the call

… the other issue, even if we do that there can be some scenarios where knowing the real device id is important

… I think we can solve the first problem by adding some more semantics

… We could have an API change if we go with the inheritance chain proposal

… could then do instanceof ...

… and then new devices are instance of a former device

… that would solve the extensibility prob but still think there is a need of pointer type

… and need to know the actual device type

… Might be a bit weird to add it to a future spec

… I think Rick is comfortable with followoing up on this later

AB: we still have 3 weeks of LC

OP: I agree we should solve this problem later

… we can't predict future extensibility

JR: there was a lot of talk about leap motion

… at W3Conf

<jrossi> https://www.leapmotion.com/

JR: this came up at edgeconf

DS: has anyone talked to them?

<slightlyoff> ...because one of the folks on the panel seemed to be a representative

JR: I have expressed interest

DS: I played with Joshua Davis' device; pretty cool

<slightlyoff> I'm gonna cede the floor on this

<slightlyoff> I owe the list email

<slightlyoff> and don't think we can make serious progress by phone

<jrossi> slightlyoff: are you just on IRC?

Should pointerId be an integer by Rick Byers

AB: Rick started a thread about the type of pointerID http://lists.w3.org/Archives/Public/public-pointer-events/2013JanMar/0146.html
... Alex mentioned this issue too http://lists.w3.org/Archives/Public/public-pointer-events/2013JanMar/0110.html.

<slightlyoff> jrossi: no, on the call, but don't want to queue in just to say "not today" = )

<slightlyoff> why?

<jrossi> just curious

JR: I'll reply to the pointerID thread

AR: why do people not want to make it opaque?

JR: I found UCs for it being an integer

… in one painting demo with multitouch, used integer to pick from a random number of colors

… perhaps not a great UC

… Do you have a proposal for opaque?

AR: think object identity should be sufficient

… can't guarnatee integer stability

… having multiple mouse isn't common

… One objection is only having one mouse and making it 0

… Don't think we want an integer for a specific device

… Breaks down when new pointer types are added

… Don't want to set bad expectations

JR: Rick mentioned some issues with using integers

… e.g. when comparing

… May need to add some more context about the integers

… Can achieve good level of interop with integers

… and make sure people's "false assumptions" are addressed

… Think this would be a problem for our impl

… if we had to switch to opaque types

… I am willing to consider it, but would prefer to keep this integer

… and to add some more information and context

… One pain point is supporting touch events which used integer

AR: given that, I think it would be ok if using integers was fleshed out better

… that would be better than creating an interop problem

JR: do you have a proposal?

AR: opening move is to write down the IE behavior

… then we have something to discuss

DS: is part of your rationale pattern searching?

AR: if integers, it permits indexing to arrays

… and that works ok if the impl moves through the integer space in a reasonable way

… but if a different impl moves through the integer space differently, there will be interop problems

… Don't want confusion for the app devs

… if the id is overloaded

DS: so, need to define the semantics of the integer e.g. before/after

AR: need to make sure impls handle integers the same way

OP: I prefer a random behavior

AR: and that would meet my opapue requirement

<jrossi> Roughly speaking, IE10 reserves 1 for mouse. Then 2+ values are used for other inputs. With each newly recognized pointer, the ID is increased. But there's some max at which we wrap back around to 2. I'd have to check with the Windows kernel folks for clarity.

Click and contextmenu events by Rick Byers

AB: Rick started a thread about click and contextmenu events in http://lists.w3.org/Archives/Public/public-pointer-events/2013JanMar/0151.html
... Jacob replied http://lists.w3.org/Archives/Public/public-pointer-events/2013JanMar/0152.html.
... is this just a matter of adding a bit of explanatory text?

<shepazu> (I think I agree with Alex about pointerId, though I want to think about it)

JR: there is some explanatory text that needs to be added

… there is already some related information

… I think it is safe to add the extra text

… The other issue is Ricking looking for a defn of "click"

… There is a defn in D3E

… Not sure if he missed it; just sent today

… This could be no change or just some additional non-normative text

<slightlyoff> still on the call = )

<slightlyoff> ES6 maps make this go away

SG: re pointerId, whatever we do we need to make sure it is compatible with JS objects

<slightlyoff> you can have arbitrary keys

<jrossi> when ES6 is interoperable :-)

… we cannot assume everyone is using ES6

Testing Pointer Events v1 spec

AB: there was some discussion on the list http://lists.w3.org/Archives/Public/public-pointer-events/2013JanMar/0157.html.

<jrossi> what about polyfills? integers will be easier if trying to polyfill older browsers, I think.

<scott_gonzalez> By compatible with JS objects, I mean specifically that the pointerId can be used as a unique key in an object.

AB: any comments re the proposed directory structure?

JR: looks good

AB: any comments re how to identify tests as manual vs. automated?

<scott_gonzalez> If we do use objects to represent pointerId, a custom toString() which returns a unique value would be fine.

DS: we should use metadata for auto/manual

… I should bring in Tobie Langel, W3C's test lead

AB: ok, let's talk about how to schedule that

<smaug> er, what is the command

AB: any comments about the need for test assertions e.g. http://www.w3.org/2010/webevents/wiki/TestAssertions? Any volunteers to lead or contribute?

<smaug> thanks

AB: any volunteers?

CC: I can help get that started

AB: excellent!

… thanks Cathy!

Any other Business

AB: does anyone have any implementation status to share?

DS: I think Jacob did an excellent job on PE @ W3Conf!

… it is available on youtube

<shepazu> http://www.youtube.com/watch?v=SCfVn4JY5yk

DS: I will write a blog re PE for webplatform.org

<jrossi> http://docs.webplatform.org/wiki/concepts/PointerEvents

… Msft has some folks creating PE materials

… they have an overview plus reference material

<jrossi> Shorcut url: http://bit.ly/pointerdoc

DS: they will ask this group for feedback

… Thanks Microsoft for making that happen!

JR: I talked to a lot of devs at the conf

… about 4 other talks mentioned Pointer Events

… and that's pretty cool

DS: yes, lots of interest

<slightlyoff> am not

AB: re next call, we have 3 more weeks of LC review

… given that, perhaps we skip next week and next call is March 12

JR: assume we want to record some Resolutions and don't want to get too far behind

… so it may sense to have a call next week

AB: good point; let's you and I chat at the end of the week re if a call on March 5 makes sense

JR: sounds good

AB: thanks everyone for joining. Meeting adjourned

- DRAFT -

Pointer Events WG Voice Conference

26 Feb 2013

Attendees

Contents