WebApps WG f2f Meeting

01 May 2012


See also: IRC log


Art_Barstow, Charles_McCathieNevile, Josh_Soref, Dimitri_Glazkov, Arnaud_Braud, Adam_Klein, Soonbo_Han, MikeSmith, Paul_Kinlan, Philippe_Le_Hegaret, Travis_Leithead, Anne_VanKesteren, Odin_HortheOmdal, Russell_Berkoff(Samsung), Rafael_Weinstein, Tony_Ross, Kris_Krueger_(krisk), Eric_Uhrhane, Vincent_Scheib, Bryan_Sullivan, Glenn_Adams_(glenn), Dan_Druta, Magnus_Olsson, Jonas_Sicking, Doug_Schepers_(shepazu), Tantek_Celik, Ted_OConnor_(hober), Greg_Billock, Adrian_Bateman, Wonsuk_Lee, Paul_Cotton, Alex_Komoroske, Ojan_Vafai, Yosuke_Funahashi, Ryosuke_Niwa
ArtB, chaals


<ArtB> Scribe: Josh_Soref

<ArtB> ScribeNick: timeless

Welcome and Agenda

ArtB: When you do introductions, please indicate if you are not a WG member

plh: as the charter has been reupped, most people are not WG members

chaals: if you were a WG member and haven't reupped, please nag your AC rep

ArtB: Josh_Soref is a fantastic scribe, he works for RIM
... RIM is not a member

<plh> WG participation status

ArtB: when you speak for the first time, please introduce yourself

chaals: when Josh_Soref says stop, you have to stop, because you'll be lost otherwise
... I'm chaals, Opera, I'm a chair

ArtB: I'm ArtB, from Nokia, I'm a chair

Paul_Kinlan: I'm PaulKinlan from Google, registered as observer, now member

ericu: I'm ericu from Google, I'm a member

glenn: glenn, Cox, member

DanD: Dan Druta, AT&T, member

Arnaud: Arnaud Braud, France Telecom, member

bryan: Bryan Sullivan, AT&T, member

Russell_Berkoff: Russell Berkoff, Samsung, Observer

aklein: Adam Klein, Google, Observer

rafaelw: Rafael Weinstein, Google, Observer

tross: Tony Ross, Microsoft, Member

rniwa: Ryosuke Niwa, Google, Member

MikeSmith: Mike Smith, W3C Team, Member
... the first one to rejoin

PaulC: Paul Cotton, Microsoft, Chair of HTML WG, your host

anne: Anne, Opera, Member

odinho: Odin Horthe Omdal, Opera, Member

Travis: Travis Leithead, Microsoft, Member

shan: Soonbo Han, LG Electronics, just joined [and was dropped by recharter]

ArtB: there's an Action for someone to bug ACs for rejoins

<chaals> ACTION: chaals to bug AC reps of ex-members to re-join after new charter [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action01]

<trackbot> Created ACTION-643 - Bug AC reps of ex-members to re-join after new charter [on Charles McCathieNevile - due 2012-05-08].

magnus: Magnus Olsson, Ericsson, Member (need to rejoin)

krisk: Kris K, Microsoft, Member

plh: Philipe Le Hegaret, W3C Team, Member

scheib: Vincent Scheib, Google, Member

dglazkov: Dimitri Glazkov, Google, Member

<MikeSmith> trackbot, start meeting

<trackbot> Meeting: Web Applications Working Group Teleconference

<trackbot> Date: 01 May 2012

<magnus> present Magnus_Olsson (magnus)


ArtB: We always preallocate an item or two
... and then figure out the rest as we meet
... we have a couple of topics
... we had penciled Intents for 1-2pm
... James_Hawkins was going to manage that

PaulC: The preallocated name badges were to help the secretary
... just register at the desk
... if you have problems, let me know

ArtB: Here is the list of potential topics
... most of them I added
... (in alphabetical order)
... and then dglazkov added components
... and bryan added server sent events

<anne> http://www.w3.org/2008/webapps/wiki/May2012F2FMeeting

<anne> ^^ meeting agenda

<MikeSmith> agenda: http://www.w3.org/2008/webapps/wiki/May2012F2FMeeting

ArtB: WebAppSec has CORS on its agenda for tomorrow morning
... they had allocated half an hour for LC CORS

<MikeSmith> Jonas Sicking has entered the fray

ArtB: 9:45-10:15

chaals: How many people are interested in CORS?

[ Quite a few hands rise ]

chaals: does anyone object to bringing them in here?

[ No objections ]

ACTION ArtB to talk to WebAppSec about a joint slot

<trackbot> Created ACTION-644 - Talk to WebAppSec about a joint slot for CORS LC [on Arthur Barstow - due 2012-05-08].

chaals: Welcome sicking

sicking: Jonas_Sicking, I'm the late Jonas Sicking, Mozilla. Not The Late Jonas Sicking, just late

chaals: Any other topics not in the wiki?

scheib: I spoke briefly with ArtB
... I'm the editor of the Pointer Lock specification
... I'm new to editing
... it's just been added to the charter

ArtB: I think it would be useful for new specs that have been added
... that people are starting to implement

chaals: I might put looking at the Charter/Schedule/New Specs
... either at the beginning. or at the end
... any preference?

anne: it might be good to put them at the beginning

bryan: I have a conflict for 11-1

chaals: we won't put push there
... going around the room

<MikeSmith> list of specs is at http://www.w3.org/2008/webapps/wiki/PubStatus#API_Specifications

chaals: we've got less gaps here (today), than there (tomorrow)

dglazkov: Shadow DOM, HTML Templates

chaals: I'll put bryan (Push/SMS) to 2:30-3 (today)
... and Web Components for 11:15-12:30 (today)

dglazkov: my items is procedural ... gauging temperature

chaals: IndexedDB

ericu: we have a request from someone from Google who can't be here today
... can it be tomorrow?

chaals: Yes we can

ArtB: I'd like a slot for Hixie 's 4 CRs
... where we are, can we get someone to fill in the gaps
... how do we manage future work
... v. getting to rec

chaals: "Hixie's hand-me-downs" 11:30-12:30 (tomorrow)

Travis: 10 minutes for DOM3 events/DOM4 from that slot?

chaals: is that going to be short?

anne: we had the longer one last time

Travis: it should be short

anne: we need 15 minutes for Full Screen
... ArtB mentioned that

chaals: the Late Douglass Schepers
... people who have not introduced yourselves
... please introduced yourselves

shepazu: Doug Schepers, W3C Team Contact, Member

tantek: Tantek Celik, Mozilla, Observer

hober: Ted O'Connor, Apple, Member

anne: Is the Stream API that's an extension to XHR going anywhere?
... is the editor here?

gbillock: Greg Billock, Google, Observer

MikeSmith: please put the IME API

chaals: I'll try to leave space for breaks
... how many people have read the new charter?

[ ~5 hands ]

chaals: 4 of us were lying

shepazu: I don't think chaals read it, and he wrote it

bryan: the link on the webapps page is to the old charter

<anne> charter: http://www.w3.org/2012/webapps/charter/

Josh_Soref: the main webapps page is unusable

bryan: the w3c pages don't work well on iPads

shepazu: Action bryan to buy me an iPad

chaals: Action bryan to buy everyone an iPad

bryan: the style sheet is generally bad

chaals: who's driving the screen?

ArtB: I am

[ ArtB projects PubStatus ]

Charter/Pub Status

<MikeSmith> new WebApps charter

ArtB: we have over 700 people subscribed to our list
... of those, only 30-40 people are really active
... I like to keep pub status accurate

glenn: I have a comment to shepazu

<plh> Agenda for today

glenn: it might be helpful to say XHR subsumes ....

shepazu: can we make emendations?

plh: nope

chaals: CORS
... we'll look at this tomorrow

anne: I don't see how it's a plan

[ the label for CORS says "LC Period ends 1-May-2012" ]

anne: but the statement is accurate
... there have been no comments raised
... there was one "we should design this differently"
... there was a comment about making it more performant on mobiles
... that was related to caching

chaals: do you expect a second version

anne: if we tinker with caching, then we'd need a second version
... there's also From-Origin (the opposite of CORS)
... there are long term plans re: merging CORS + fetching

shepazu: you should talk about that in the slot

chaals: If you're speaking, you need to speak loud and to the center of the room

anne: we can't fix all the bugs

ArtB: so move to CR?

anne: there have been no LC comments

chaals: we expect to move to CR in two weeks?
... Clipboard APIs and Events
... Halvord Stein is not here
... is anyone following that closely enough?

[ no ]

ArtB: do we know implementation status?

<tantek> One request for the new WebApps charter (starting July 1 2012 presumably) - please switch to using the W3C wiki (instead of webapps WG-only wiki) : http://www.w3.org/wiki/Webapps/

anne: there's implementations
... but they have differences

rniwa: depending on platforms, there are variations
... there are issues involving determining Same-Origin
... affecting what can/should be stripped
... it might be needed

chaals: so that's work in progress

anne: there's an attribute for secure usage?

chaals: CORS - testing
... and test facilitator, and test suite?

odinho: me

chaals: what's the status of the test suite?

odinho: for the test suite
... I've been reading through the tests that are there
... I've incorporated the things that are missing into Opera's Test Suite
... but I haven't gotten entirely through the WebKit tests

chaals: and that hasn't been sent back to the group

krisk: tests that are submitted are a wide range
... we should go through them

sicking: we have a couple of tests that are pretty big
... but they won't run anywhere else (they use "yield")
... would you like us to submit those
... a lot of the tests are expressed as data
... you could write a new wrapper around it

odinho: I've looked at it

chaals: I'm trying to get a bird's eye view
... summary: odinho is looking at it others are working on it
... is there a test coordinator for Clipboard APIs

rniwa: I don't think so
... how would we test it?

<plh> I think it would be good for Mozilla to submit what they have, and we figure out in the longer run how to modify them

rniwa: it can't be from the web page
... so it has to be manual

shepazu: so you define manual tests

anne: there's a WATIR framework

<anne> Watir

chaals: don't sign up to do something if you don't have the bandwidth for it
... DOM4

anne: the Plan statement (for DOM4) isn't quite correct
... at some point we'll add new features
... better event registration
... extending ClassList
... varadic? arguments

chaals: do we push DOM4 through and start DOM5
... what's the rush to get DOM4 finished

anne: you could push DOM4 through and work on DOM5

<Zakim> MikeSmith, you wanted to ask if somebody wants to give update about plans for Quota API spec

anne: but we don't have a way to manage forks (maintaining DOM4 and working on DOM5)

plh: we can't link to an unstable thing from a spec

chaals: that discussion is about w3c process
... that's out of scope for this WG

anne: I know there are people that want it
... but I have limited bandwidth
... we could publish dom4 now
... it's way better than dom3

chaals: Adrian Bateman, Microsoft, Member

Travis: only the WebPerf WG has requested to link to DOM4

plh: a bunch of specs want to

rniwa: there's demand to deprecate DOM Mutation events (DOM3)
... I think Mozilla is planning to unprefix the replacement

chaals: it sounds like it would be good for the chairs to find someone with the bandwidth to branch of DOM4 and stabilize it
... is that someone standing up to volunteer?
... thank you very much Tantek

ACTION ArtB to find someone to branch DOM4 and publish

<trackbot> Created ACTION-645 - Find someone to branch DOM4 and publish [on Arthur Barstow - due 2012-05-08].

anne: if you make the CR requirements loose, we can do it fairly quickly

ArtB: is anyone interested in helping with that task?

[ Silence ]

chaals: don't worry anne, we'll come back and ask you again
... until you come up with the right answer, which is yes

Travis: ArtB, please show PubStatus wiki page

[ ArtB captures need to fork DOM4 for stable+publishing ]

<anne> WebApps Pub Status (on screen)

Travis: bugzilla database is the prime spot for tracking (DOM3 Events)
... I think we should issue another LCWD

chaals: DOM Parsing + Serialization

anne: the HTML WG might or might not work on it

chaals: it's in our charter

PaulC: the CfC for HTMLWG:ISSUE-198 closes today

anne: in particular, if it closes, it will be forked from the html
... and someone from microsoft will publish it

chaals: despite the fact that it's in our charter, we don't know if it will happen in our group
... is that right paulc?
... my sense was that we would do it in our group

anne: no, they wanted it in the html WG

PaulC: I'd have to do the research
... I don't think HTMLWG:issue-198 speaks to where it would be done

ACTION chaals to talk to paulc about where Parsing+Serialization work is done

<trackbot> Created ACTION-646 - Talk to paulc about where Parsing+Serialization work is done [on Charles McCathieNevile - due 2012-05-08].

chaals: Element Traversal is DONE
... File API

sicking: the pub status for File API looks right
... we can possibly do it in Q2

chaals: do we expect Q3
... let's say we expect it in Q3
... directories and systems

ericu: that's all correct

chaals: From-Origin Header

anne: I don't think there's been much uptake
... drop it, I guess
... I've addressed all the comments
... there haven't been other comments
... I don't think anyone implemented it
... the idea was to prevent people from using CORS in places for which it wasn't quite intended
... but they started doing that anyway

chaals: so that has no one to take it forward
... does anyone want it?
... it's up for grabs

anne: I'm happy to continue editing it
... but if no one is going to implement it, then there's not much point

chaals: let's start a CfC to publish it as a note
... if that doesn't shake anyone out, then park it as a note

<ArtB> ACTION: Art start a CfC to stop work on From-Origin spec [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action02]

<trackbot> Created ACTION-647 - Start a CfC to stop work on From-Origin spec [on Arthur Barstow - due 2012-05-08].

ACTION ArtB to start CfC to publish From-Origin as a note

<trackbot> Created ACTION-648 - Start CfC to publish From-Origin as a note [on Arthur Barstow - due 2012-05-08].

bryan: I understand technically what it was intended to do
... and I understand it was a good idea
... but I'd like to understand how CORS stands if we don't have From-Origin

chaals: Full Screen
... do we have a test coordinator?

anne: no

chaals: ok, so we need one

WonSuk: WonSuk Lee, Samsung, Member

plh: From-Origin is in the WebAppSec Charter
... so we should talk to them

ArtB: I didn't think it was a joint item

plh: we can talk to them tomorrow

adrianba: Fullscreen...
... is it two specs?
... there's a CSS bit

tantek: it will be managed together

ArtB: how close is it to somewhere?

<ArtB> ACTION: Art start a CfC to publish a FPWD of Fullscreen spec; coordinate with CSS WG [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action03]

<trackbot> Created ACTION-649 - Start a CfC to publish a FPWD of Fullscreen spec; coordinate with CSS WG [on Arthur Barstow - due 2012-05-08].

chaals: we expect a FPWD this Q
... Gamepad

scheib: The Gamepad editor is Scott Graham, from Google
... the draft has been stable for the last little while
... chrome is behind a flag
... I believe Firefox is soon to ship without a flag
... I don't see anything blocking

plh: publish as LC?

shepazu: FPLC?
... it's kind of funny

chaals: you can do that
... full screen might do the same

<ArtB> ACTION: Art start CfC for FPWD + LCWD of Gamepad spec [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action04]

<trackbot> Created ACTION-650 - Start CfC for FPWD + LCWD of Gamepad spec [on Arthur Barstow - due 2012-05-08].

shepazu: why don't we have a session to do them

chaals: Indexed DB
... we have a test suite
... it's on the agenda
... anything to say?

<anne> fwiw, Gamepad is not ready for LC

<anne> at least https://dvcs.w3.org/hg/gamepad/raw-file/default/gamepad.html is not

sicking: I don't think there's much to do

chaals: IME?
... MikeSmith ?

<anne> e.g. GamepadEvent does not inherit from Event at the moment and does not define a constructor

MikeSmith: do you need a Test Facilitator?

chaals: yes, thanks

MikeSmith: I'm happy to do it

chaals: we need a FPWD

<ArtB> ACTION: Art start CfC to publish FPWD of IME spec [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action05]

<trackbot> Created ACTION-651 - Start CfC to publish FPWD of IME spec [on Arthur Barstow - due 2012-05-08].

chaals: anyone following Java bindings for WebIDL?

Travis: I don't know anyone doing it

chaals: I used to
... pointer lock?

scheib: I'm the editor

chaals: do you have a test facilitator?

scheib: I don't know

chaals: it's someone who commits to getting tests

<ArtB> ACTION: Art start CfC for Pointer spec [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action06]

<trackbot> Created ACTION-652 - Start CfC for Pointer spec [on Arthur Barstow - due 2012-05-08].

chaals: you can do it yourself

scheib: I'll probably do it myself
... I'm not sure of the timeline

chaals: Progress Events
... waiting on implementations

anne: there's a test suite
... Ms2ger wrote tests that end up testing WebIDL
... which people get wrong
... the test suite doesn't test dispatch
... just the interface

chaals: status?

anne: when is Opera going to pass the test suite?

chaals: Quota

MikeSmith: I thought Kinuko Yasuda was working on it

chaals: and that doesn't have a test facilitator
... looks like we need a lot of test facilitators

ArtB: yeah, a lot of holes

chaals: selectors
... it's waiting on me
... it's waiting on WebIDL
... as WebIDL is going to CR
... I think Selectors can go to PR
... the test facilitator should be me

plh: we should have a link to the interop report

chaals: expect an advancement to PR to Q2
... then it blocks again until WebIDL moves forward
... Server Sent Events

ArtB: we published a LC last week
... we have 3 weeks
... I think there was a comment last week

glenn: there was a comment about infinite reconnects

chaals: we have comments
... I think everyone's had the same issue

ArtB: the only tests I know of are Opera's
... can you submit them?

odinho: yeah, we can

chaals: Shadow DOM

dglazkov: been working on spec
... we have a test suite
... dominic has been doing them
... the spec is fairly stable
... I was going to ask about moving it to WD

chaals: the procedure for moving to FPWD
... or LC
... is: as an editor, you write to the chairs and say "i think we're ready"

<ArtB> ACTION: Art start a CfC to publish a FPWD of Shadow DOM [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action07]

<trackbot> Created ACTION-653 - Start a CfC to publish a FPWD of Shadow DOM [on Arthur Barstow - due 2012-05-08].

chaals: we write to the group asking for CfC

ArtB: the thing about FPWD is that it starts a call for IP exclusions
... it's good for the feature set to be defined at a high level
... so the IPR guys can look at that

dglazkov: we're well past it

chaals: in that case, we should [already] have a FPWD
... and we'll do that with you

dglazkov: it's well pas that point

chaals: URL

MikeSmith: looking at anne

anne: I'm working on encodings
... abarth was editing, then mike

ArtB: there's a warning from abarth

MikeSmith: we need to look through the tests
... next month we can look at it
... we could publish a FPWD now
... I can put it together

<ArtB> ACTION: Art start a CfC for FPWD of URL spec (Mike to not be lead Editor but will help to drive it) [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action08]

<trackbot> Created ACTION-654 - Start a CfC for FPWD of URL spec (Mike to not be lead Editor but will help to drive it) [on Arthur Barstow - due 2012-05-08].

chaals: FPWD needs to lay out what the thing does, which we're at
... Screen Orientation
... aka Screen Lock
... view orientation

ArtB: Mounir is working on it

[ plh, the Frenchman, properly pronounces his name, and asks how there could be a problem pronouncing it ]

sicking: I don't know

ACTION ArtB to follow up with mounir about status of Screen Orientation

<trackbot> Created ACTION-655 - Follow up with mounir about status of Screen Orientation [on Arthur Barstow - due 2012-05-08].

chaals: WebIDL
... Travis, look good?

Travis: I am the test facilitator, but I haven't facilitated

chaals: Web Intents

gbillock: we probably need a test facilitator
... I'll sign up for that

Travis: we need a FPWD

gbillock: we'll talk about that this afternoon

chaals: Web Messaging

ArtB: in CR

chaals: as of this morning
... that's PostMessage

ArtB: according to caniuse.com, it has the most deployment

chaals: but no tests

shepazu: I don't think this is the right room to draw them from

<ArtB> ACTION: barstow find a Test Facilitator for Web Messaging CR [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action09]

<trackbot> Created ACTION-656 - Find a Test Facilitator for Web Messaging CR [on Arthur Barstow - due 2012-05-08].

chaals: Web Sockets
... we need to finish CR/Test suite

krisk: MikeSmith helped get a server up
... I think MikeSmith 's going to update one module
... but it seems to be going along pretty well
... tests are pretty complete

ArtB: so MikeSmith will update the module

chaals: run the Test Suite, ask for PR

ArtB: are you aware of implementations that pass everything?

krisk: we're pretty close

anne: there's a problem in Web Sockets relating to Isolated Surrogates
... the spec requires throwing
... but preference is to replace
... I don't think it's tested by the test suite

glenn: there was discussion on isolated surrogates in public-script-coord

anne: it's related, but it's not the same
... and it won't change
... spec requires throwing
... most want not throwing

adrianba: I thought we threw

anne: for consistency with XHR which doesn't throw

Josh_Soref: and web authors won't expect it to throw

krisk: I think we should talk about this in our Hixie specs slot

chaals: Web Storage

ArtB: I think there's a late DOM4 change
... which blocks Web Storage
... does anyone implement that?

krisk: I don't think anyone does yet
... it's definitely blocked on that

ArtB: Yikes,

krisk: we should talk about that in the Hixie specs slot

chaals: Web Workers

ArtB: CR today

Travis: someone doing that should work on Web Messaging, since they're intertwined

anne: Web Workers has feedback that may require going back to LC

chaals: that's right
... into that slot too
... XBL2
... anyone love that enough to follow up?

ArtB: wait for sicking

chaals: my impression is that it's going to be parked

anne: I think smaug is the only person who cares

chaals: XHR

anne: 2 doesn't exist
... I wrote a test suite once
... but no one cared
... I tried to find someone, and odinho ...

odinho: I had an intern

chaals: making an intern isn't a good idea
... since they disappear
... we got a request from Mozilla when we rechartered
... to look at web app packaging
... sort of a JSON version of Widget Packaging
... and we have a potential draft starting point
... do you, tantek, have any further idea on its status?

tantek: is this Manifests?

chaals: yes

shepazu: yes
... and do you know who that is?

tantek: I think that was Michael Hanson
... what's the input you are requesting

chaals: it's in our charter
... Mozilla has a spec and someone supposedly into it
... do they have someone to do the work
... and you can say I don't know

tantek: I don't know

chaals: the answer is "we don't know"

ACTION shepazu to contact dbaron (Mozilla AC), cc tantek

<trackbot> Created ACTION-657 - Contact dbaron (Mozilla AC), cc tantek [on Doug Schepers - due 2012-05-08].

[ Break for 15 minutes ]

Web Components

chaals: sicking wasn't here
... XBL2 should be parked as a WG Note

sicking: if things go south, can we bring it back?

chaals: yes, it's in the charter

plh: is there a lot of work?

shepazu: do we do a CfC?

chaals: I volunteer to update the status of the document

<ArtB> ACTION: Barstow start CfC to create a WG Note for XBL2 (and Chaals will do the work) [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action10]

<trackbot> Created ACTION-658 - Start CfC to create a WG Note for XBL2 (and Chaals will do the work) [on Arthur Barstow - due 2012-05-08].

chaals: where are we?

dglazkov: lots of work has been done since last TPAC
... the main feedback at TPAC
... was we brought a lot of stuff
... but it was a bag of goods
... rather than a coherent whole
... we needed a declarative form
... where is the spec
... confinement/isolation
... lightweight/functional
... with the help of ArtB and shepazu, we got things we needed
... it takes more work to get a component in WebKit Bugzilla

<shepazu> Web Components Explained

<dglazkov> Shadow DOM ED

dglazkov: we talked to a lot of people
... I tried to come up with as solid of a spec as I could
... simultaneously we developed this in WebKit
... behind a flag, and only available in Developer builds
... I don't want a repeat of WebSQL
... this helped inform ourselves about things
... it helped flush out things
... the basis of the spec was the XBL2 part
... there has been a lot of things added
... a lot of that is precision of Shadow DOM
... htmly things
... guided by our implementation
... today the spec is in pretty good shape
... we have a small bug list

<dglazkov> Shadow DOM Bug Tree

dglazkov: some are small things, "not MUSTy enough"
... there's one (largish) addition we're contemplating
... bug 15818
... I also worked on the HTML Templates Spec
... an idea
... we have the templates element (see the Explainer doc)
... what makes it "interesting" is that it requires HTML Parser modifications
... I wrote the spec and WebKit modifications
... to see how it was received
... several people voiced Cautious Concern
... Hsivonen and Abarth
... the two parser people whose brain's we picked
... they James Graham from Opera wasn't very happy either
... there's still a need for an extra mode (?)
... the <template> tag has 2 modes
... "declare anything"
... "declare anywhere"
... we're going to drop "declare anywhere", we don't need it for Web Components
... "declare anything" we're going to keep, since it seems useful

Josh_Soref: you're going to drop anywhere, and you're keeping anything?

[ Laughter ]

dglazkov: right
... Custom Elements is the next spec in line
... I'm planning to start working on it next week
... I spent the last couple of weeks researching the problem space
... I wrote a poly-fill
... if you have Shadow DOM

<dglazkov> Polyfill (using Web Components)

dglazkov: in Custom Elements, one of the new thing is fictional syntax
... these items aren't controversial
... loading definitions of components
... which is a big issue
... don't want Synchronous
... but Asynchronous has issues: When am I a component/When am I unknown?
... instantiating a Component
... has interesting effects
... maybe I'd like to be able to drop into user script
... but if I'm instantiating from Parser, that maybe isn't a good idea
... more mundane issue
... custom elements are DOM Objects
... with an extended prototype chain
... I don't know how to spec this
... since it creates a dependency on ECMAScript

anne: what exactly?

dglazkov: Custom Elements extend the Prototype Chain

<plh> partial interface?

dglazkov: I don't want to create a dependency on ECMAScript

anne: why is creating a dependency on ECMAScript a problem?

plh: it seems like you're creating a partial interface

dglazkov: right, but it's arbitrary

anne: you should talk to heycam
... he'll probably say that you have to define it yourself in prose
... I'll look into it next week, after this session
... another thing, relating to elements
... we came to TPAC with custom tags: x-slider
... there was much grievance
... we switched to use elements
... it's a magical element, you can only set it once
... button is fancy button
... during instantiation, you have to specify it
... Eric Myer, of myerweb

<dglazkov> http://meyerweb.com/eric/thoughts/2012/04/10/element-customization/

anne: I want to bring it up, I'm feeling very ambivalent
... I'd like to figure out who would be the right person or forum
... I posted to webapps
... and crickets chimed in

tantek: isn't this more HTML WG than WebApps WG?

dglazkov: that's what I'd like to know

shepazu: about inheriting from Button / Slider / Calendar
... there's been talk in the past about having pseudo elements
... say for CSS
... say for the slider's thumb

dglazkov: we looked at the CSS variables spec
... the spec says CSS variables inherit into Shadow DOM

shepazu: that calls out the need for pseudo elements

dglazkov: with CSS variables, you don't

shepazu: I don't understand yet

<dglazkov> http://dvcs.w3.org/hg/webcomponents/raw-file/tip/samples/widget-theming.html

dglazkov: I know the MS guys did pseudo elements
... and we have them in WebKit
... and we hate them
... what browsers use pseudo elements to style bits of things
... I think you use pseudo classes

sicking: what do we use to style the placeholder
... or input elements?

tantek: it's a pseudo class

dglazkov: in any declarative paradigm
... if you're saying button, or div is shelf
... you're defining a subclass
... and when you instantiate it
... you don't say, it's a shelf, oh, it's also a div

shepazu: could there be another thing other than localname?

dglazkov: I don't want to mess with

tantek: could you consider defining it as a mixin rather than a subclass?

dglazkov: that's decorators

tantek: why not have everything be a mixin?

dglazkov: when you're dealing with everything as an API
... you want to ensure things are always the same
... you don't want a style recalculation to cause your object to lose its decorator/API?

tantek: that's done through CSS?

dglazkov: well, decorators are done through CSS
... and then there's the moving it out of the tree

tantek: well like the class= attribute

dglazkov: but then you can have "spooky action at a distance"
... if you change the class name
... what happens to its state?

tantek: that invariant could be maintained

dglazkov: I think that's possible
... when the developer of a component
... relies on it to be a button
... if you want to have multiple things as a tree
... that's definitely possible

sicking: it introduces complexity
... roughly Multiple Inheritance in C++
... it's very powerful, but very complicated

dglazkov: [currently] it's just extending a prototype chain
... moving down the chain

tantek: it seems like reinventing Java Class Hierarchies

dglazkov: it's not reinventing
... just naturalizing JS inheritance into the DOM

shepazu: in SVG we have the USE ELEMENT
... you have a use element, you reference an element
... and you get another instance
... but you can add attributes to the copy
... so you can have a plain star
... and then style one copy to green
... or red
... at TPAC, you said "No"
... is that still the answer?

dglazkov: I think having the shadow tree with separate style
... has dragons

shepazu: so we could reuse it?

dglazkov: you will lose some of the invariants that the SVG spec provides
... but the way Shadow DOM is defined

shepazu: I think all that SVG needs to keep
... is the way to style each instance separately

dglazkov: that's possible today
... SVG uses Shadow DOM in a very limited way
... it doesn't have insertion points
... if you want to extend to that
... it's OMG

ArtB: from a procedural perspective
... we agreed to a CfC for Shadow DOM
... what about Template Spec?

dglazkov: I think Template Spec, as it is right now, we're going to kill it
... and we'll pursue it in HTML
... in "Custom Elements"

rafaelw: I think that's fine

ArtB: so we're not going to publish this

dglazkov: we're probably not going to publish it

rafaelw: one of the more salient issues of the template element spec
... is two things
... what mechanism creates inertness
... and where do those elements reside?
... on the ML, there was a propose that they be lifted out
... there was an item about it being objectionable
... if we can sort out that
... I think that's the most useful+controversial part
... if we can get consensus on that, I think we can get progress

[ Time Check ]

anne: hsivonen, abarth, jgraham are not here
... hober is here
... basically you lock out XHTML uses of templates

hober: all things being equal, we shouldn't introduce more divergence between HTML and XHTML

dglazkov: we have a Mexican Standoff
... between should we hurt XHTML
... vs. should we introduce something very non performant

<MikeSmith> cough TAG cough

ojan: Ojan, Google, Member

<shepazu> Alex Komoroske

komoroske: Alex Komoroske, Google, Observer

anne: elements inserted based on the template element
... you don't want them to be in
... because they cost resources
... and are exposed by QuerySelectAll/etc.

dglazkov: can we modify the XHTMLParser?

anne: I have a draft that tries to modify XHTML parsing
... but it hasn't ...

dglazkov: can we CfC dropping XML?

[ laughter ]

chaals: there's a proposal to make XML a kinder gentler beast

anne: there's a big leap for moving things into a detached dom tree
... it's cool and works for me

<MikeSmith> need a magic namespace

anne: but I don't think it would fly for others

tross: technically, it's inserted into the tree
... and then removed before anyone looks

anne: the people who will care is the TAG
... and they want a document served as HTML or XHTML to behave the same

PaulC: More specifically, the Director cares

chaals: changing XML is like changing the W3C Patent Policy
... but it isn't written in Stone
... it's on a wiki somewhere

anne: we can say "we want to do this in html"
... we don't think it will work in xml

PaulC: make a comment on the html WG's document

anne: that's done, there's a bug
... but that won't get TAG attention until it ships

ojan: no one has an alternative proposal that's technically feasible
... every other proposal has serious technical problems

anne: if you don't address hiding from DOM Query

ojan: and every future API that might do a network request or live action request
... needs to be template aware

anne: in effect everyone needs to be aware

shepazu: can we introduce the feature and say "does not work with xml"
... and let them solve it?

anne: we already have that, it's called <noscript>

dglazkov: we could also say we require an esoteric changes

sicking: if I were to do this in Gecko
... I wouldn't touch Expat
... I'd change how the tree constructor handles events from Expat
... I don't think we need to violate XML
... wasn't there a proposal to stick things into <script> tags?

dglazkov: there was

sicking: although that also doesn't work in XML

chaals: we can say "hey world, we're going to upset your apple cart/orchard"
... and see if they care

MikeSmith: +1

anne: you definitely violate the spirit

chaals: there's no question that it makes a mess

tantek: if you're using XML, can't you use XSLT?

dglazkov: resolution: we'll try to spec it as "doesn't work in XML"

tantek: I don't think it's Apple Specific

[ Laughter ]

ArtB: dglazkov, have you thought about publishing the Explainer?

dglazkov: I thought about it
... but it seems like a sequencing issue

chaals: it makes sense to do it

dglazkov: I can reformat it
... update it (for Shadow DOM)
... and then publish

shepazu: I can help

dglazkov: if you guys have time
... please dig into Shadow DOM and help me eliminate non MUSTy stuff

anne: there might be a lot of those things

<ArtB> ACTION: barstow start a CfC to publish a FPWD of Web Components Explainer (when an ED with TR template is available) [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action11]

<trackbot> Created ACTION-659 - Start a CfC to publish a FPWD of Web Components Explainer (when an ED with TR template is available) [on Arthur Barstow - due 2012-05-08].

dglazkov: I do spend a lot of time staring at the spec
... but it's hard for the person who wrote something to see its faults
... any more questions?

chaals: thank you dglazkov

[ Applause ]

chaals: let's have an hour for lunch. Resume at 1:30pm

Web Intents

<dglazkov> http://dvcs.w3.org/hg/webcomponents/raw-file/tip/spec/shadow/index.html

Paul_Kinlan: Hi, turns out I'm a member
... as of a couple of hours ago
... we're going to talk about web intents
... I want to give you a demo
... I don't know how much you know about what we're trying to achieve

[ We try to get projector projecting ]

[ Lights go out ]

[ chaals: Nope, that's not the one ]

Paul_Kinlan: there are a couple of UCs where it's very hard to build integrations
... with third parties
... the whole point is that even though we have widgets
... there's no way to make integrations
... the biggest common action is Share
... the next is Bookmark
... but things people want to do:
... Edit Documents, Pick resources
... we want to make this easy
... let someone pick something from their cloud storage

[ Projector had temporarily done something positive ]

[ Projector failed ]

[ adrianba: I just pressed random things until it worked ]

[ shepazu: that's how they do most things ]

[ Laughter ]

[ chaals: alright, do the interpretive dance ]

Paul_Kinlan: there are a couple of common actions
... that we think are core to the web
... users do common things:
... share data
... save physical data to things (like Drop Box)
... they pick data from things (Word Document, Image, Video)
... they could pick from Flickr, Drop Box, YouTube
... one of the things I was going to show (in the demo) was Imagemator
... what we would see on the screen is a big button that says "Choose image"
... the browser knows which services you use
... the demo would let you pick from Picasa
... Picasa doesn't use Intents, but it has a public API which lets you do it
... You can do server to server work
... but we're starting to see purely client side applications
... a lot of their functionality is built on the client
... when you have applications sharing lots of data (Video, ...)
... the data might be local to your network
... or proximate to your network
... we'd like to let these two applications talk directly
... like a bridge
... we have demos that do both
... need a network, or client side resolution
... where we literally process a blob
... the demo itself doesn't do much work
... it finds a service that does editing
... if the browser doesn't have a service for a thing
... the browser can use an indexing service (store, search engine)
... to discover a service
... In the demo, you'd press edit
... you don't have anything installed
... the Chrome Store would be searched

<anne> link to the demo?

Paul_Kinlan: you'd pick Picnik

s|anne -> http://demos.webintents.org/ Web Intents Demos|-> http://demos.webintents.org/ Web Intents Demos|

<anne> Imagemator

DanD: have you looked into a scenario
... where the application developer wants to choose a certain instance
... say I'm a photo sharing service and I want to choose Picnik
... I want to do it in a way that makes sense
... not to choose a default intent
... but a specific case

Paul_Kinlan: We've talked about that in the TF
... an "Explicit Intent"
... say you're photoshop.com
... You want to be open to discovery
... however you've got specific integration with

DanD: and the user should be able to override in the end?

gbillock: explicit intents, it's unclear whether they will be overridable
... explicit intents let web content make the picker
... and letting web developers use Web Intents for internal RPC
... the way that you could bring up a browser guaranteed redress proof UI
... is interesting
... we're hoping with experimentation we'll figure that out

chaals: the answer is that it should be overridable

shepazu: it should be up to the UA

chaals: if you have a local installed application
... it should work
... say you don't want to use photoshop.com
... you want to use Photoshop

gbillock: that's totally within scope
... we definitely want to be able to build a bridge between web apps and local apps
... for some embedders like OSs
... like Android
... there ought to be a way to create a mapping
... or Windows 8's "Contracts"
... you shouldn't be able to just go from a web app to Photoshop
... but also from Photoshop to say save to your dropbox

Paul_Kinlan: we also want to be able to do viewing
... it should be easy to do Open With

gbillock: currently the spec is focused on what you do for the Web Page
... there's language to say that this isn't the only way
... saying that there should be a local execution model
... but that's left up to the UA

chaals: how do you go with AppCache

Paul_Kinlan: we've done a lot of experimentation with AppCache
... we've experimented with RPC/RPH
... it's hard to get things to work with AppCache'd content
... most people use a query string with RPH
... but Intents uses something different, so it could work

magnus: you said you could have a UA
... that could download it using a search engine
... what happens while it's being retrieved

Paul_Kinlan: the implementation in Chrome
... does the query using the web store (http)
... the API itself is Async
... the UA pops up the picker
... but the page isn't blocked
... if you have no networking
... then there might be no options for the user
... but how that works is up to the UA
... and because it's Async, that shouldn't affect the page

gbillock: the idea that a user might be trapped with no options
... is definitely unappealing to developers
... one possibility is to let clients query to see if things are installed
... but that leads to fingerprinting
... that's a weak supercookie
... instead the direction we're trying to go
... is to let client applications provide fallback suggestions
... that the UA can use if the picker would otherwise be empty
... instead of being empty, you might see DropBox or whatever
... our current experimental implementation uses the chrome web store
... so they have to be installable
... the end state we'd want to get to
... is to have a web for web pages to identify themselves as services
... we've been discussing that in the HTML WG
... do we have an <intent> tag
... or ...
... It looks like Hixie is most favorable to having an <intent> tag
... but combining RPC, RPH, <intent> together
... so they'd look the same for users
... giving us both Imperative and Declarative
... and the same User Facing appearance

anne: what Hixie said was quite reasonable
... that still doesn't say how you identify an app

gbillock: Gmail would say use RPH for mailto:
... and register <intent>

anne: on my web page, I have a contact form
... and I have a send me an email link

<gbillock> Web Intents specification

gbillock: if you look at 3.1
... there'd be a services parameter
... in chrome
... the picker is a list of optional services
... the top having items the user has used
... possibly it would query the store

anne: if the developer provides URLs
... what do you show?
... not just the URL?

gbillock: no, the page title + favicon, probably
... or if we've processed it, something else

chaals: I want to go back to overrides
... the UC will come from Accessibility
... if you made a request for an Explicit Intent
... it should be possible to pick something else
... if you pass off text, then anything can
... but if you pass word97 documents
... then there are some other things that can handle it

shepazu: obviously, if I have something, I can describe it
... is there any other way to give information to the user?

gbillock: in the picker itself?
... the client presents the initial messaging to the user

shepazu: is there a way to give a human-readable description of the requested action?

gbillock: the UA has the complete Intent call bundled up
... what the action is
... what the type is
... any extra data
... the UA has to use that
... it can definitely customize itself
... to say "which of these services do you want to use to edit a contact"
... we got a bit of feedback from UI people to provide per action wording

[ We have a projected Cosmos ]

Web Intents Demo

gbillock: Selection refers to the picker
... since the UA is in charge of that
... the UA can be arbitrarily sophisticated in terms of coaching the User

<chaals> Imagemator

Paul_Kinlan: this is Imagemator

[ Clicks Choose Image ]

Paul_Kinlan: these are the user's services
... and these are the store services
... I'll use ... CloudFilePicker.com
... this is Picasa
... it isn't direct, it's via the Picasa API

[ Picks a face with two phones pasted ]

[ Laughter ]

[ Clicks Edit ]

scribe: I'll pick Mememator, I haven't installed it
... this has no server to server logic
... eventually this will work offline
... I'll pick Inspirationmator

[ Enters Practice Demos; They work ]

Paul_Kinlan: The UA passes the data around

glenn: does this pass data around retain tainting?

Paul_Kinlan: not in this case
... in here, the canvas isn't tainting
... I want to show two actions "Share blob" and "Share page"
... Web Intents can handle both

adrianba: who decides what to be shared?
... in the Windows 8 contracts, we publish different options
... the link, the link with metadata, the html

shepazu: like clipboard

Paul_Kinlan: like clipboard
... right now, the application invokes one type
... saying I'm invoking the Image
... the link would be the physical image, and not a reference

gbillock: there are two strings
... for match making
... the actions and the type
... the actions must match exactly
... and types must match, or if they're mime types must overlap
... if Twitter knows to share Images, Links, or Videos
... then it would register for 3 distinct things
... so you get a footprint over all the things you understand
... that's our theory right now

Paul_Kinlan: the client application says it will do one thing
... your application will say it can support three types of data
... we might need to change it so you can offer one of two things as a request

adrianba: there's a problem where you have multiple datatypes with precedence
... but it seems that like now the onus is on the user right now
... I know that Twitter can take: page link, page link+title

Paul_Kinlan: I think the onus is on the Client app to pick sensible types

adrianba: as a user of the source app
... I have to know which button to pick to trigger to the destination app I have

Paul_Kinlan: right now, our apps have one definitive type/action
... share was kind of interesting
... because very few apps share physical data
... most share data

shepazu: I agree with adrianba, share is ambiguous
... look at facebook
... at one point you only shared a link
... now it also embeds some of the content

gbillock: the thing starting the activity is the client
... and the thing performing is the service

<Zakim> shepazu, you wanted to ask about "inlining services" into a page with intents

shepazu: I think there will need to be a negotiation

gbillock: the question of how complicated the handshake should be
... is obviously
... in order for this to work, the ecosystem has to agree
... Facebook/G+ occasionally figure out what you meant
... with that in mind, we've erred on the side of no negotiation
... expand what you except

anne: I think for most user how to pick a service will be complicated enough

gbillock: we decided to burden the service to enumerate what it supports

anne: maybe you should have a way for the client to offer multiple at a time
... and let the service indicate its preferred payload

Josh_Soref: this is not "Paste Special"

shepazu: the user is stupid (sarcastically)

anne: the user has better things to do

<Zakim> timeless, you wanted to talk about Share v. Save

Josh_Soref: I made a trip and tried to sign in. I could print a PDF or follow a link. I would like to decide to send it somewhere.
... you were expecting me to send it via a sharing service, but I want to save it somewhere and then use that to do my sharing.

gbillock: You want to be able to translate the intents?

Josh_Soref: I am saying they are the same thing

Paul_Kinlan: I don't have an answer - there are different things that people expect from what they see.
... I don't think we don't want to fire two intents, or people will end up publishing washing lists of services that do everything.

Josh_Soref: fallback is to have translator intents. Doable and I want to make it easy - but I see share and save as the same.
... I can print to my device, rather than on paper. It is really a save, but as far as the computer is concerned it is a print.

Paul_Kinlan: Share was a broadcast, save was putting it somewhere. I can see the mental models behind this, but we have to work on this

gbillock: The API doesn't spell out the verbs. It is an invocation of delivery leaving things open for usage to coalesce.
... reason common verbs are useful is that they give a way to develop a good expectation to agree on what you are trying to do.
... There are edge cases which are hard to think about - should a kindle support print and share and save?
... we're waiting to see what happens with usage - what emergent verbs there are.

komoroske: we are waiting to get feedback.

shepazu: In your demo you open a tab for the events. It might be interesting to be able to load a service inline on a page...

Paul_Kinlan: we have two dispositions. All these demos use new tab, for transitory implementation motivations (bugs)
... there is an inline disposition that should be able to do that.
... let's you see the context, it is relatively spoofable, it is an area we have been wary of.
... we weren't confident that we could make it secure.

shepazu: can't you have a UI option where the user gets to choose how it appears?
... e.g. in maps I want to have something within a page.

Paul_Kinlan: we want to explore it but haven't.

gbillock: the obstacle is that the service has to provide an iframeable interface, which is subject to attack and we haven't figured out how to solve that yet.
... we are expecting a proposal from someone so we will see what happens.

<Zakim> adrianba, you wanted to talk about example of sharing a page

adrianba: wanted to give an example from windows. We have contracts, and share contract is one of them. The browser supports the idea of sharing a page. User decides to share a page.
... browser is a client in web intents terms. Can share link, a link+metadata, or HTML snippet.
... when I choose share, windows finds services that supports one of those formats. Twitter app might take links+data, a bookmark does something similar, email might use the full HTML snippet, ...
... we allow any service that responds to a type to appear. User doesn't have to think about the options.
... sounds like your model is the service say it can take one of those three.

Paul_Kinlan: We ahve a model where you can share a link. Once we have that we can go fetch more detail, and put it in the metadata pat of the intent.
... you have a link plus extra metadata. Have to think about service applications - they can ignore data, read it if it is there.
... not all clients will share all data. In Android services don't populate metadata consistently.
... using name as a URL was based on describing a particular experience. Tell people how you do it, what to populate the data with. Was going to be a lot looser on definition, with people using URL as value for where the information will come from.
... both client and server would choose what they send/receive.

adrianba: feels a lot less predictable about what the user is going to receive
... what is the URL for, what can I do with that?

anne: doesn't the service also register types it accepts? I think you have the same system.

adrianba: if I publish a URL of something with an image, do I mean the page or the image on it?

anne: I think it makes more sense to send both.

gbillock: one way to do both is an option we discarded (can reconsider). To integrate with types, we intend that it be possible to match a microdata type with complete schema capability...
... contact might be name+phone, or might have a lot more data there.
... idea is that user has a mental image of the service they are using.
... and so builds expectations of what is going to happen.
... There is flexibility in terms of how much payload is available to fill in for the service.
... weakness and strength.
... if your phone accepts a contact with no phone number., that seems wrong

shepazu: Could you use this across multiple modalities?

gbillock: We envisage the user agent being able to do things like use NFC to send stuff...

DanD: Who is in control of selecting the directory of services?

gbillock: User Agent. Services the user has installed that meet the required intent.

DanD: who provides the list of options for what to install?
... on mine it is meaningful - it gives me information about where the suggestions are coming from.
... good this is under control of the browser for sense of trust, user needs to know where the browser is going.
... side effect of that control limits discovery of other services which may be an issue.
... (vendor lock-in...)

gbillock: Think client will be able to attach suggestions.

DanD: more appropriate for app developer to recommend the directory, rather than having the user search the web. But you have to put the destination for searching into the user experience

gbillock: this is a stock UI for inline installs. if there are suggestions from the client side they show differently. Being able to attribute stuff comprehensibly matters...

DanD: deja vu here - this is uddi/wsdl/etc. again...
... there were some good developments done there, so looking for the lessons there is a good idea.

<Zakim> tantek, you wanted to ask how broad is the scope of intents and cross-application services, e.g. some examples discussed seem similar to OpenDoc/OLE, especially in local client-app

tantek: demos are awesome. scope is broader than I had understood. How broad is the scope intended to be?

gbillock: spec is 'how pages invoke intents or get them delivered'
... leaves it up to the browser

tantek: web apps, client apps, installed web apps are all in scope?
... that page has all the ability of HTML to send data anywhere, on the web or locally.
... including native apps?

gbillock: in principle yes. we haven't done that yet, but it is in scope.

komoroske: scoped to websites, but could do this if there is demand.

Paul_Kinlan: DAP is interested in this. We have been focused on webapp interactions. UA can provide a bridge to add native apps.

tantek: do you know about OpenDoc and OLE?

<tantek> OpenDoc

<tantek> OLE

tantek: systems for applications doing this. Have you looked at that?

gbillock: nope.

tantek: there is IPR in that area that should be looked at.
... (I know because I did some of it)
... I felt I had an obligation to make that disclosure

<Zakim> shepazu, you wanted to ask about a site registering itself as a service

shepazu: I am on Flickr, it wants to tell me it can be a picker service. Is there something that lets them put something on their page so I can register it when I go there?

gbillock: yep.
... right now we have experimental stuff, but yes we want to be able to do that through declarative syntax for the page.

shepazu: if I share stuff with Twitter, can I make that a default rather than picking every time?

gbillock: spec leaves that to user agent, we expect that to be possible.

Josh_Soref: having something is important to avoid security issues - you don't want a spamming site to get your Twitter

shepazu: there should be a user involvement to make sure

gbillock: there is.

<Zakim> timeless, you wanted to note WAI concerns and Portability/Modality concerns

Josh_Soref: if the client page is making a request and can force a directory that doesn't work for my device, or have an accessibility requirement for specialist services, or want a different language,
... the client might not have the right answer for the user.

DanD: we already have scripts that pick stuff...

Josh_Soref: right, but they are not necessarily useful for a new device.

DanD: agree there may be an incompatibility. Would rather have app developer test and verify than have the user agent assume the thing will work.

[ Break until 3:30 ]

<plh> Current group participants

Push SMS

chaals: we have an item in the Charter for Server Sent Events
... the two things people have come up with are Push notification stuff
... and a notification that can wake up / remotely start an app (web page)
... I'll hand the floor to bryan

<chaals> [Yosuke Funahashi introduces himself - co-chair of TV/Web IG]

yosuke: Yosuke Funahashi, co-chair of TV/Web IG

bryan: I've taken the UCs and broken them into a set of

<ArtB> Push UCs

bryan: more discrete things which I'll call proto requirements/ideas
... there's a link to a w3-ified draft from within OMA

<ArtB> EventSource Push (Draft)

bryan: it doesn't address all of the requirements
... I've built this, and have a demo (which I won't try to show today)
... and there will be a social network demo called "Mobile Social Networking"
... at XXX
... we noticed that XMPP connections burn battery real fast
... I want to get notifications of things that are really asynchronous
... e.g. an auction watcher
... doesn't want to keep an application open
... until recently, you couldn't run a browser in the background on mobile devices
... the next UC is a WebRTC client
... the phone application/dialer
... it doesn't take over the screen until it needs to
... we need some way to register for wake up events
... we were looking for a way that was more seamless
... @ TPAC:WebApps last year
... there was a request that things not be so specific
... my proposal was based on my experience w/ SMS/OMA Push
... but we need to create a mapping between text-eventstream and these other things
... I ran into an issue involving blank lines
... Maybe we end up building on a processing model

[ bryan is reading through the Push "Derived Requirements" section ]

bryan: there needs to be a way to provide filters
... the ability to deliver information to a web app before it shows a UI
... my draft proposal incorporates CORS
... to apply the browser security model

<Zakim> chaals, you wanted to ask where we go with this now...

chaals: when this came into the charter
... I'm not sure if the people who wanted it are here

[ sicking raises his hand ]

chaals: you have a draft idea of a spec

bryan: that may be fairly localized in application

chaals: do you have this in web apps space?

bryan: not yet

chaals: so you're planning to edit this

bryan: I could definitely support that
... I'm looking for expert input

chaals: next step is to put it into w3 space
... and put it in the list of work items

bryan: I'm hoping to have a conversation on things

chaals: sure, but you start with an ED

bryan: sure

sicking: "We"
... (loosely)

<sicking> http://jbalogh.me/2012/01/30/push-notifications/

sicking: also have a draft proposal
... it doesn't cover everything from the proposal
... we could add things
... an application can say "i want to be able to send push notifications to the browser"
... the user agent allows the user to authorize that
... and if authorized, a URL is made available to the application
... and then it can use whichever applicable means to send messages back to the UA/page
... it should be integratable with Apple's push protocol
... currently you can't deliver that message to a particular page
... it shows up on the screen
... but when the user clicks on it, it goes to a certain page
... we could let pages say they don't want things on screen

bryan: it sounds like OMAPush
... service indication
... (pre 2000)
... a text message and a URL
... that wasn't directed to an application
... We added a way for an application to listen directly
... the key to OMAPush is that it uses Tokenization
... in 4 SMS payloads, you get up to 2k of content
... which isn't achievable without tokenization

chaals: so it sounds like we have two sort of half starting points
... going into the same direction
... so the action is to look at them together

bryan: I can look at mozilla's draft
... there's interest in integrating Apple's Push Notification
... and C2DM
... (Google's)
... that's where it stands

ACTION bryan to look at proposals and start editing

<trackbot> Created ACTION-660 - Look at proposals and start editing [on Bryan Sullivan - due 2012-05-08].

magnus: the proposal is to extend Server Sent Events
... with event streams
... but you're not limiting to that

chaals: we're not limiting to that

bryan: it may, but I found hoops, WebSockets may be better

DanD: I'm a member of WebRTC
... this came up as a requirement
... it got escalated to WebApps
... we did some analysis

<tantek> <aside> chaals, follow-up from your question about Application Manifest, we (Mozilla) do have someone working on a spec, and are iterating in public with intent to submit to Web Apps WG for inclusion/publication: http://mozilla.github.com/webapps-spec/ cc:sheapzu,sicking </aside>

DanD: it's nice to have
... but there are emerging technologies which will make it a necessary feature

<chaals> [reply to your aside: Cool. We chairs are waiting for that :) ]

bryan: in this draft, you'll see some text examples
... I'll send a link to the github which has a demo

Arnaud: is there a speed requirement?

bryan: I haven't seen any request for a service delivery deadline
... things tend to happen within a second or two

Arnaud: if you use SMS as a bearer
... it can be slow

sicking: Apple's has no promise of delivery at all

bryan: it's best effort

File API

chaals: where is Mr. Arun?

sicking: this is a side project for arun
... he hasn't been able to work on this for a while
... he did a spurt of editing LC feedback into the spec
... I have to work off memory of the outstanding issues
... the big one is One-Time-Only
... and revoking

<chaals> File API bugz

sicking: I think there was something else

adrianba: there was the Close thing

sicking: I think Close has the same problem space as Revoke

adrianba: how concrete do we have to be
... and how interoperable do we need to be
... in IE we have a behavior where something may be cached in the decoded Image cache

sicking: I suspect we'll want to define those things
... I suspect we'll be done with the File spec
... before we have those cases done for Images
... I suspect that long term we'll want to and require behaviors

<Zakim> chaals, you wanted to generalise Adrian's question onto tomorrow's agenda

chaals: I would like to take this point out of this discussion and put on the agenda tomorrow whether we can a stable version of the spec
... that gives a useful stable reference
... while we work forward

adrianba: there's a difference
... between is it valuable to have specs that are roughly stable and useful
... and there's a part of a spec where there are so many variations based on underlying platforms
... saying for those things maybe we don't have to specify them maybe ever

sicking: I suspect we'll want to define this
... I suspect for image cache, we probably have the same issue
... and if you hadn't brought it up
... we may not have tested it

adrianba: web developers care
... and they care when it breaks them

ericu: they care if we underspecify it
... and it works in one browser and breaks in other browsers

sicking: it'll affect every place that uses URLs, and every place that reads out of blobs/files
... there are certain things we should define
... there are going to be lots of things we're going to miss
... some of these things should be specified outside the File API spec
... some I suspect we'll get to eventually

shepazu: if there are several contentious issues
... and some that will be tricky to do
... maybe we should bring on an additional editor

chaals: this isn't an editor issue

sicking: it isn't an issue of the File API spec
... it's up to the other specifications to accept a hand off

chaals: should we put out a call for a second editor

ericu: oh, we have a second editor

[ sicking raises hand ]

shepazu: you're just very busy

sicking: there's a very small amount of this that will go into the File API
... for One-Load-Only
... do we revoke at first access or at end of microtask
... the other is...
... if you start loading, and then you revoke
... should that load continue
... I think on the second one, I don't think we've gotten feedback from you, Microsoft

adrianba: oh, I can give you feedback:
... once we've started, it's very hard to stop

sicking: ok, so once a load has started it should finish

adrianba: for Close
... if you're doing File Reader on a Blob
... the point of calling Close
... is to say you really want to let go of the resources

sicking: I considered them to be the same
... but we can keep them as separate
... we should figure out
... the first thing is ArrayBuffer v. ArrayBufferView
... I dislike the topic enough that I haven't followed the discussion

Josh_Soref: +1

sicking: I suspect that we should be using ArrayBufferView

adrianba: we can't do that soon

sicking: I'm happy to leave it as an OR

adrianba: I agree that it seems like it should be the right thing
... by the time we could change
... ECMA TC39 could progress

sicking: I suspect that even if TC39 does something, it'll be called as ArrayBufferView or subclassed as that
... my feeling is we do ArrayBufferView now
... and if something new is added, we can add it later

adrianba: do we always know
... when you use these things
... can we reliably feature detect support for these?

sicking: the Blob constructor is hard to detect

anne: you could just try

chaals: you guys drink beer tonight and solve the problems

anne: you're going to have the need for feature detection all the time
... it hasn't been a real problem in practice

adrianba: I think these are relatively new features that haven't been detected

anne: you need browser sniffing anyway

adrianba: for incremental things
... for response-type in XHR, that's a good feature detect
... set and retrieve
... some things we're adding will be harder for feature detection
... there's an envelope thing with Must understand
... and things which aren't

chaals: I really do mean: have a beer tonight, talk about this

PaulC: I'd like a beer too

anne: we can discuss it tomorrow
... feature detection is the agenda item

ericu: File Writer, Locking
... sicking has a new proposal
... but we need to discuss it on the list

[ Break until 4:30 ]


<ArtB> IME UCs and Requirements

MikeSmith: I worked with Kenji and Hironori
... the main cases where IMEs are important are Japan and China
... and to some extent Korea

chaals: and Vietnam

MikeSmith: since you don't have 10,000 keys on your keyboard
... you type on your keyboard, it goes into a buffer
... and gets converted
... into Kanji
... you don't want an IME to interfere with Games
... or other things
... similar to Screen Orientation/Pointer Lock
... interactively typing and getting suggestions from a web application
... like Google Suggest
... completing against things in a database in real time
... while you're completing against a database, you're potentially also completing against the IME

Josh_Soref: most mobile devices also have IMEs for word completion for Latin languages

MikeSmith: Interaction
... some people want feature compatibility with other runtimes
... Flash has the ability to interact with the System IME

Josh_Soref: System IMEs are BUGGY AND INSECURE

MikeSmith: as a game developer you can use this
... some people want to be able to provide a web based IME
... if you want to create a complete branded and consistent UE
... then if your application includes text input
... and you want to control IME behavior in your application
... then you want to be able to brand and style that

yosuke: a lot of systems don't have IMEs installed
... and users don't know / can't install them
... so a web site might want to provide that
... installing that may require privileges

MikeSmith: I don't know why we didn't put that one

Josh_Soref: google provides that for Translate for Hebrew

chaals: Yandex does that for Cyrillic

MikeSmith: Hixie didn't think this was the right approach

tantek: there's an existing CSS property ime-mode
... from IE5/Firefox
... that should address your Games case
... it's in CSS3 UI LC
... it's at risk
... there's a simple property there
... if there are UCs that are easy to add to them
... the spec is being locked down

<tantek> http://www.w3.org/TR/2012/WD-css3-ui-20120117/#ime-mode

rniwa: there's no concept of IME on/off
... when you switch languages/layouts
... I don't think this new API addresses that either
... maybe there's a way to include that

anne: given that web pages already make their own UIs
... it'd be helpful if there was an explanation as to why something is needed


chaals: when you implement it, what happens in practice is it doesn't work

<rniwa> Josh_Soref: ah, ok. hober: thanks

rniwa: there's some interest in creating SVG editors
... you want to be able to type things into the SVG
... and that isn't compatible with Content Editable

Josh_Soref: some "system" IMEs are buggy and pushing insecure content into them
... is just as dangerous as pushing data to font engines
... = nice root exploits

<chaals> MikeSmith: We're about ready for a FPWD, the spec is reasonably advanced. Hopefully sometime this month.


<tantek> perhaps consider an informative reference to CSS3-UI for the 'ime-mode' property

MikeSmith: the URL Spec

<ArtB> URL spec

MikeSmith: the API part
... if you're going to expose URL information
... then you want a way to parse them
... the part before this is the algorithm for parsing
... there's a definition of what a URL is
... it isn't defined anywhere
... the first two parts started in the html spec
... but there isn't anything specific

<tantek> I've done some research on what different specs call the different parts of URLs: http://tantek.com/2011/238/b1/many-ways-slice-url-name-pieces

Agenda Bashing

chaals: there's a gap for people to wake up
... CORS
... D3E/DOM4
... Testing
... Versions/Stabilize
... -- these points here have dragons
... Feature detection
... -- anne + adrianba 's item
... [ their action was to drink beer ]
... Meeting Planning

PaulC: when will you do that?
... I'd like to try to be here
... I'd like to have our TPAC plans straight

chaals: yes, that being our next meeting
... anything else people want to put in our next meeting?

glenn: where are we drinking beer tonight?

chaals: that's later in today's agenda
... 9:45-10:15 CORS w/ WebAppSec
... HTML Stuff
... Index DB
... [ real item ]
... Full screen

anne: 10 minutes

ArtB: what's HTML?

krisk: Hixie specs (Sockets, Workers, ...)

chaals: Lunch
... Feature detection/stability
... Testing
... Meetings
... - wrap up + beer

<krisk> Tied House Brewery & Cafe 954 Villa Street Mountain View, CA 94041 (650) 965-2739

chaals: thanks all

[ Adjourned ]

RRSAgent: make minutes

trackbot, end meeting

Summary of Action Items

[NEW] ACTION: Art start a CfC for FPWD of URL spec (Mike to not be lead Editor but will help to drive it) [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action08]
[NEW] ACTION: Art start a CfC to publish a FPWD of Fullscreen spec; coordinate with CSS WG [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action03]
[NEW] ACTION: Art start a CfC to publish a FPWD of Shadow DOM [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action07]
[NEW] ACTION: Art start a CfC to stop work on From-Origin spec [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action02]
[NEW] ACTION: Art start CfC for FPWD + LCWD of Gamepad spec [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action04]
[NEW] ACTION: Art start CfC for Pointer spec [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action06]
[NEW] ACTION: Art start CfC to publish FPWD of IME spec [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action05]
[NEW] ACTION: barstow find a Test Facilitator for Web Messaging CR [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action09]
[NEW] ACTION: barstow start a CfC to publish a FPWD of Web Components Explainer (when an ED with TR template is available) [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action11]
[NEW] ACTION: Barstow start CfC to create a WG Note for XBL2 (and Chaals will do the work) [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action10]
[NEW] ACTION: chaals to bug AC reps of ex-members to re-join after new charter [recorded in http://www.w3.org/2012/05/01-webapps-minutes.html#action01]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2009-03-02 03:52:20 $