Technical Architecture Group Teleconference

18 Jul 2013

See also: IRC log


Dan Appelquist, Marcos Cáceres, Sergei Konstantinov, Yves Lafon, Peter Linss, Noah Mendelsohn, Alex Russell, Henry S. Thompson, Anne van Kesteren (IRC only)
Tim Berners-Lee, Yehuda Katz, Jeni Tennison
Dan Appelquist
Henry S. Thompson



DKA: Welcome to new member Sergei Konstantinov (Yandex)

<twirl> Hi, everyone :)

TC39 / TPAC Update and Next Steps

DKA: Update from Philippe le Hegaret on this

<slightlyoff> no discussion in tc39 that I'm aware of

DKA: So the question is how we move the discussion from a private conversation to engage all of TC39
... to make good on our promise of making TPAC a welcoming place for TC39

AR: I'll bring this up in person at the TC39 meeting next week

DKA: Before our next call?

AR: Yes

Review current work including new work items and spec reviews

DKA: Round robin?

HST: +1 to round robin

DKA: MC, anything happening?

MC: Not sure, given uncertainty about how my participation will go forward

PL: I have been working on github mirroring and so on

<slightlyoff> plinss: exciting!

PL: I'll give more info when I'm ready to go public

DKA: Progress on website redesign?

PL: No, will set up a repo and get that out

NM: I mentioned the HTTP2.0 which you might want to look at at some point

HST: I've spent the last week or so collecting statistics about http usage in conjunction with http 1.1 review.

HST: About to do a TAG blog post with some numbers.

HST: Highlight is - I think HTTPbis needs to look very carefully at what it has to say about content negotiation.

HST: I've looked at 100m http requests and only a couple of hundred 300 responses - responses supposed to be given in 'reactive' content negotiation.

<slightlyoff> ht, can you characterize the services that the requests are from in the post?

<annevk> (300 is a kind of a useless response code. Nothing happens.)

<slightlyoff> ht, are they from many services? a few?

<slightlyoff> ht, that sort of thing

AR, That's harder, privacy issues mean in most cases I don't have request info.

<annevk> I don't have anything. I'm been traveling and otherwise occupied.

AR: I spoke with Chris Wilson about Web Audio, which I've made an initial stab at writing up
... Plan to work on that with YK and also do a draft review

DKA: I reviewed your Web Audio document, I think it looks good

AR: There's an important architectural issue which isn't addressed in my document yet: shared mutable data which is viewable from multiple concurrent javascript thread
... That breaks a core invariant of the JS processing model
... I have a proposal on how to avoid this, but I'm not quite ready to propose it until I research things more
... I hope they can sidestep the problem via my approach, but it's going to be tricky

DKA: Timeline?

AR: I'm ready for feedback ASAP on the document as it stands, wrt both content and format/style -- is this the kind of way we should be feeding into WGs?

DKA: The variation between formal and informal in the document did strike me

AR: Put in an issue or pull requests for some changes

DKA: Style isn't the main point
... What else do we have in github in the spec review area. . .
... Any other issues?
... I added the Push API to the spec review list

<slightlyoff> have the chairs thought about a plan for how to work through the backlog in a structured way?

DKA: I reached out to Charles and Art of Web Apps to try to begin coordination between us and Web Apps
... Certainly we should plan for joint meeting at TPAC
... We may ask Art to join us in Cambridge at the Sept. F2F
... But we need to move sooner on key topics -- Charles is keen to see the Promises discussion moving soon

AR: Wrt the Status pages which the W3C maintains, is there a process we have / should have to monitor that?

<slightlyoff> right, makes sense

AR: I'm happy to work in a "put out fires" informal way, but do we need more?

DKA: I was expecting to use the spec review list for this

<Yves_> Some groups have a dashboard, like webapps. Otherwise the WG page and their publication list.

AR: That's fine, but how do we keep confident that we keep up to date on what's new and might need to go on there

MC: I agree, is ad hoc monitoring sufficient?
... So, recently, a Mediastream Image Capture spec came up, do we need to put it on the list -- i.e., do we treat all new specs equally?

DKA: We could have a semi-automatic process, where we periodically go through new specs and vote them in or out

AR: Maybe just auto-generate a bug on the spec review list for new pubs

DKA: Sure -- propose something by email?

AR: Will do

PL: YL did take an action to try to automate this, at the F2F

DKA: But YL is currently blocked on that with hardware pblms

<slightlyoff> oh, right

<slightlyoff> totally forgot about that...sorry Yves_

DKA: Ashok Malhotra has sent email about a draft candidate for a TAG blog post based on our Publishing and Linking work—once we have a look at it, I expect we will say yes

PL: We still have a decision to make on whether we have a 90-minute meeting on a regular basis. . .

<slightlyoff> I thought the proposal was to leave it open

<slightlyoff> not to commit to a longer meeting every 2 weeks

<Noah> I think you should send a signal as to when TAG members should try to hold > 30 mins free, and when not.

<Noah> My concern is that when you do want a longer call, you'll find conflicts

DKA: I thought the plan was a) have a 30-minute standup every week, and to have the possibility of deciding to extend some calls for technical discussion if we have something queued up

<slightlyoff> +1 to what DKA jsut said

<twirl> +1

<slightlyoff> agree with that

<Yves_> Note that inviting spec editors and not only chairs might be intetesting

PL: Note the YK has a conflict post 30-mins every other week, so we need to be aware of that

<dka> Proposed RESOLUTION: have a 30-minute standup every week, and to have the possibility of deciding to extend some calls for technical discussion if we have something queued up, taking into account members' scheduling issues and providing the group adequate notice.

<slightlyoff> +1

<Yves_> +1

DKA: Notice would normally be given by the Monday before

RESOLUTION: Have a 30-minute standup every week, and to have the possibility of deciding to extend some calls for technical discussion if we have something queued up, taking into account members' scheduling issues and providing the group adequate notice (ideally by the Monday of that week EOD).

PL: Maybe we end each standup with a call for tech. topic for the next week

HST gives regrets for next 6 weeks :-(

DKA: What about discussing http2.0 next week?
... But HST won't be here -- take it to email?

<slightlyoff> ht can't make next week = (

<slightlyoff> +1

HST: You can talk about it w/o me, but sure, let's start by email

DKA: Adjourned

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2013-07-20 15:38:42 $