Technical Architecture Group Teleconference -- 12 Dec 2013

<dka> trackbot, start meeting

<trackbot> Date: 12 December 2013

<slightlyoff> on the way

<JeniT> Yves, are you joining us?

Call: TAG telcon

<scribe> ScribeNick: ht

<scribe> Scribe: Henry S. Thompson

F2F logistics; F2F planning; Developer meet-up status; Spec reviews: Push API, Animation, Web Crypto

London (7-9 January) F2F planning

DKA: JT, all ready for us?

<slightlyoff> Zakim: aaaa is me

<slightlyoff> I'd be happy to go out

JT: Yes, room booked, local arrangements page up and sent out

<slightlyoff> I feel like arranging lunch has been a burden for organizers so far

JT: Go out for lunch, or have things laid on?

TBL: Two catered in, one out

JT: Will do
... Light breakfast too?

DKA: Yes, please
... Developer meet-up -- looking for venue
... Hoping to get Google campus, not confirmed
... Fallback is GSMA
... This is all for evening of the 7th
... I'll put this up on Eventbrite
... Dinner out on evening of the 8th

<dka> http://www.w3.org/wiki/TAG/Planning/2014-01-F2F

Agenda page for January F2F

<slightlyoff> oh good

<slightlyoff> can we schedule time to discuss the layering of Web Components at the F2F?

DKA: Invited Dominique Hazael Massieux to join us to talk about [commissioning?]

<slightlyoff> I think it'd be an important thing to review

DKA: Hoping to invite Phil Archer to pick up on the Data activity

<slightlyoff> sure

<slightlyoff> also, the JSON thing

<slightlyoff> it's coming! really!

<timbl> :)

<slightlyoff> do we want to talk about the JSON thing today?

<slightlyoff> hrm...

DKA: Yes, good, we'll put the Web Components topic on

<slightlyoff> do we want to put any of the HTTP2 stuff on the agenda? is there anything new there?

<slightlyoff> ht: hah, happy to oblige = )

TBL: Timing for each day?

DKA: 0830 for 0900 start
... End at 1730

<slightlyoff> that's jetlag friendly = )

<slightlyoff> early mornings are jetlag friendly

<slightlyoff> if you're coming from the west

<timbl> no

<slightlyoff> guess it depends if you prefer evening or morning flghts? ;-)

Extensible Web Submit

<timbl> early morning are not friends if you come from the west, they feel horrible

<JeniT> add capability URLs to agenda

<timbl> early mornings are good if you are coming from the E

<slightlyoff> and we're looking for a bay-area venue?

<slightlyoff> do we care that it's in SF proper?

<slightlyoff> or not?

<slightlyoff> ok

DKA: Still looking for a venue -- date is fixed for 4 April
... Downtown, or OK further afield
... Suggestions welcome

<slightlyoff> Google is hard...but I can look

Spec reviews

<slightlyoff> I'm prepared to talk about push API

<dka> https://github.com/w3ctag/spec-reviews/blob/c218958ce2c9b655282bd1ebdbe93e8902b423e1/2013/08/Push%20API.md

<dka> https://github.com/w3ctag/spec-reviews/blob/master/2013/11/WebCrypto.md

<slightlyoff> ah, yes

DKA: Push API, Web Animations, Web Crypto

<dka> https://github.com/twirl/spec-reviews/blob/ceb61bdc7969f730475d23e338c31b05071b9ee6/2013/10/Web%20Animations.md

<slightlyoff> my apologies that I haen't finished something coherent for crypto

Push API

Spec Reviews

Push APIs

<slightlyoff> https://dvcs.w3.org/hg/push/raw-file/tip/index.html

AR: Push API has some good parts and some less good parts
... I'm not happy with the layering aspects

<slightlyoff> https://dvcs.w3.org/hg/push/raw-file/tip/index.html#navigator-interface

AR: Current spec says Navigator gets a Manager attached to it

<slightlyoff> function registerPush() {

<slightlyoff> navigator.push.register().then(

<slightlyoff> function(value) {

<slightlyoff> pushRegistrations.push(value);

<slightlyoff> registerWithAppServer(value);

<slightlyoff> },

<slightlyoff> function(reason) {alert("darn! "+ reason );}

<slightlyoff> );

<slightlyoff> }

AR: Good promise-based registration
... Why is this different from ??
... Why register for multiple servers

<slightlyoff> if (navigator.hasPendingMessages("push")) {

<slightlyoff> // do any special pre-processing before push messages are delivered

<slightlyoff> }

<slightlyoff> last_version = null; // initialize last message version

<slightlyoff> navigator.setMessageHandler("push", gotPush);

<slightlyoff> navigator.setMessageHandler("push-register", reRegisterPush);

<dka> AR: worst part of API is this.

<dka> AR: setmessagehandler seems to suggest you'll be pushing a javascript message without a specific context.

<dka> ... who wants this - in a world where if you close the tab you can't get push messages. Or is the idea that when the tab is reawakened you now get the push messages? And if so why are these not events just sent to the document?

<dka> AR: The use case isn't particularly clear.

AR: These should just be events, instead of messages
... I think a good chunk of this needs to be events-based, rather than message delivery

SK: I agree, I had the same difficulty figuring out the context/background assumptions

<dka> Suggest a way forward here: Alex can you capture your thoughts into the document, then send them over to the working group chairs and spec editors...

SK: I think it's premature to send a review until these things are clear

<slightlyoff> I like this plan

DKA: So we need to fold what AR said into SK's document ASAP, then send some kind of notification of our higher-level concerns (as opposed to a detailed review) to the WG, and ask them to join us on a call soon.

<slightlyoff> +1

AR: Right, doc. as it stands is not ready for an implementer to read.

DKA: We do have one call scheduled before the end of the year

<slightlyoff> happy to draft something that that repo today to outline areas for discussion

DKA: I will try to line them up for 19 December if we can move that fast

<dka> +1

<slightlyoff> SGTM, I can be there on the 19th

SK: +1

<plinss> +1

<slightlyoff> yep, will do

DKA: First step is to get AR's comments into the feedback document

<slightlyoff> apologies for having delayed this discussion until this week

DKA: Let me know when that's done, and then I'll get in touch with the editors

Web Animation

Web Animations Review

SK: Comments, anyone?

AR: Where are they in the Process? Are they at Last Call?

PL: Still [an ordinary] WD

<slightlyoff> http://dev.w3.org/fxtf/web-animations/

<Yves> still regular WD, not even LC

DKA: SK has raised some issues in his review [see above]
... Mostly not major, but levels of abstraction could be

SK: I didn't find any major issues, I thought the spec was very good
... There are related contentious issues, and some weird behaviours are suggested

<slightlyoff> there are some real naming issues (the shed should be green!): "SeqGroup", "ParGroup", etc.

SK: It's a very complicated spec, we really need another reviewer

DKA: Can someone volunteer to spend the time

<slightlyoff> I should...I've been asked to look at it from the Google side as well

PL: I already volunteered, will do as soon as I can

<slightlyoff> at least everything has constructors = )

DKA: It would be good to finish this off by the F2F

<slightlyoff> most of the Google folks are in Sydney

DKA: And maybe have a live feedback session via IRC/Voice/...

<slightlyoff> Tab's usually PST

PL: Coming from the FX task force [?], mostly SVG input latterly
... Timezones not v. friendly

Web Crypto

DKA: AR, ready to talk about this yet?
... There's been a lot of work going on

AR: Group has a lot of challenges
... Particularly wrt scope
... Hard to manage many use-cases driving into every discussion
... Trying to decide about low- vs. high-level API
... Currently aiming for low-level
... Looking at streams vs. promises, worried about churn in the promises spec
... Question about including algorithms into discussion of APIs/implementations
... I'm not too worried about the overall API
... I think we may need to get involved wrt layering
... So consider secure vs. insecure keys
... There's a use-case challenge there which hasn't been fully recognized
... How can you talk about embedding MD5 in Javascript, when Javascript strings aren't byte arrays, for example, of layering.

<dka> Does it make sense to schedule some f2f time in London on this with the wg chair Virginie?

TBL: A well-defined coercion from string to bytes, would that help?

AR: Absolutely -- TAG could usefully push for this

TBL: From Crypto to TC39?

AR: Probably better from the TAG

<slightlyoff> +1

DKA: Virginie, WG Chair of Crypto, is in France, perhaps could come to F2F to talk about this, and on Security on the Web topic
... Other topics?

<timbl> got the dates fior ioetf london

DKA: Heads up wrt IETF in London, with STRINT workshop ahead

<dka> https://www.w3.org/2014/strint/

<timbl> March 2-7, 2014

<timbl> is ietf

DKA: Strengthening the Internet against Prevasive Monitoring
... Feeds into HTTP-2 discussion of TLS
... Adjourned

<slightlyoff> sgtm

- DRAFT -

Technical Architecture Group Teleconference

12 Dec 2013

Attendees

Contents