[CSSWG] Minutes and Resolutions Hamburg F2F 2012-05-10 Part II: Administrative, Testing

Summary:

   - The CSSWG is getting overloaded; discussed ways of scaling the WG resolution
     process to handle the higher flux of issues. Some ideas, no conclusions.

   - Tentatively scheduling F2Fs in Tucson next February and Tokyo in May.

   - RESOLVED: Send Style Attributes spec to PR once implementation reports are
               published on www.w3.org.

   - Håkon encourages people to review the new tests in the multi-col test suite.

   - Discussed separating the test approval process from the test review process,
     as we did for 2.1

====== Full minutes below ======

Administrative: Making Decisions
--------------------------------
Scribe: fantasai

   Topic: Scaling the WG resolution process to handle higher flux
   plinss: We're getting busy. Got a lot more work than we have time to do.
   plinss: Daniel and I were debating what to do about that
   plinss: we can increase length of telecons, or length/frequency of F2Fs,
           or both
   dbaron: Another suggestion is to try to resolve on more things not on
           telecons or F2Fs
   dbaron: try to resolve issues by email, the way WebApps group does
   * hober yes!!!! this!!!!
   dbaron: They post a Call for Consensus, and make decisions by email
   ?: public list or csswg
   Bert: I won't see it on the public mailing list
   plinss: too much noise on the public list
   glazou: I have conceptual problem with making all that on the mailing list
   glazou: When on conf calls we say we'll review document, make a decision
           in 2 weeks time
   glazou: Who reviewed it? One, maybe 2 persons?
   dbaron: it will work if you actually make the decision whether or not
           people reviewed it
   glazou: That's not how it work, make a decision, then months later
           someone will reraise it
   dbaron: This is one of the only WGs on core stuff that is in web browsers
           that still does synchronous decisions
   glazou: Taking HTMLWG as an example is not a good idea
   dbaron: large number of groups that switched to this model
   glazou: we don't all have to do the same thing, and this WG is still working
   dbaron: A lot of us frustrated by how we spend time in meetings
   sylvaing: we can scale using the same way
   glazou: We can try, but I don't think it will work well
   <sylvaing> I don't think dbaron was saying the WG does not work currently,
              simply that our way or working cannot scale
   szilles: I think in the cases where there is a fairly clear solution,
            concrete proposal you have week to review and raise objections
   szilles: that can easily work on email
   szilles: More concerned about kind of discussion we just had here, e.g.
            rules for overflowing
   szilles: there was alternative proposals, really hard to get out all the
            alternatives
   szilles: useful for group time, you're educating bunch of people
            simultaneously and discussing ...
   hober: group requires so much specialized knowledge to make decisions with
   hober: I have much trouble remembering details in synchronous fashion
   hober: if it's by email, I can take a day to look into it and then respond
   fantasai: ... [there's value in both email and synchronous discussions]
   fantasai: ... [could have model that's a hybrid of both]
   jdaggett: [discusses difficulties of 2am calls]
   glazou: We don't take all items to discussion on telecon, only items
           that seem to need discussion
   glazou: Given high volume, it's hard to know when a thread is stabilized
           and ready for WG resolution
   vhardy: I've seen WebApps people able to resolve issues
   vhardy: do they have a tool to support the discussion, or is it all email?
   Tab: Having something that needs discussion than mailing list isn't only
        reason it goes on agenda
   Tab: Some thing need resolution by the WG
   Tab: e.g. flexbox issues, not all of them need to be discussed F2F
   fantasai talks about having better preparation of issues before they go
            on the call
   szilles: To submit an agenda item, you have to post that summary
   dbaron: would prefer to do wiki than email
   glazou: want it archived better than on a wiki
   dbaron: problem with email is that people reply to it, and it becomes
           a thread
   dbaron: alternative proposal
   dbaron: maintain a list of proposed agenda items in a wiki, and when the
           item comes up for discussion, chairs paste that into the agenda
           email
   sylvaing: so you queue things up on the wiki, then archive the actual
             agenda on the mailing list
   szilles: if they put it on the agenda, does it get cleared from the wiki?
   fantasai: suggest that the chairs take responsibility to clear the wiki
             after it's been discussed and closed
   glazou: problem isn't this, problem is number of specs
   ...
   Tab: we have a scaling problem, we have to solve it
   Tab: we didn't have as much to talk about before, now we do, and we have
        to solve the scaling problem
   sylvaing: It would be nice that the half of animations issues that can
             be resolved on the mailing list could be done
   plinss: I don't have a problem with it in principle
   plinss: But we've tried it, and it hasn't work
   plinss: We've taken issues to mailing list before, and next week nothing
           else has happened
   plinss: My fundamental concern with trying to resolve on mailing list is
           the overwhelming amount of traffic on www-style
   plinss: Nobody can read it all
   plinss: if we say, here's the discussion and we'll resolve unless there's
           object, only 3 people will read it
   hober: call for consensus would have to be a new thread
   sylvaing: but who in here has ability to make informed decisions on
             everything we discuss (aside from dbaron)?
   Tab: you flag those thread
   plinss: Even issues that get flagged, easy to miss
   plinss: ... [anecdotes from using mailing list and missing things] ...
   Florian: People who participated in the original thread start a new
            thread tagged [call for resolution] with the summary
   glazou: you don't think contributors on www-style won't reply to those
           threads?
   dbaron: if someone is making too much nose, tell them to stop
   shane: if this is a problem, why not start another list for resolution
   shane: publicly-readable, WG-writeable
   <hober> shane++
   discussion of public vs private mailing lists
   jdaggett: let me turn this around and say, glazou and plinss who are
             objecting to this, maybe you have something in mind
   jdaggett: are you proposing more meetings, or what?
   plinss: I think short-term, we should extend telecons to 90 minutes
   plinss: pretty much every telecon we cut someone off at the end, extra
           30 minutes would help
   krit: I think it's a benefit to have 1 hr, have to concentrate on something
   hober: Then instead of talking about something for 45 minutes that should've
          been 20 minutes, would let that run to 60 minutes instead, no win
   szilles: ... modularization
   Florian: not about modularization, about breadth of CSS
   szilles: one suggestion I have is if you went to a longer telephone call,
            that you have a part of it which is general topics, and another,
            announced ahead of time, that is focused e.g. Flexbox
   szilles: and people who don't want to engage, can drop off during that call
   hober could have separate calls
   fantasai: if you want to talk about text, maybe do it not at 2am Japan time
   * stearns thinks this is important to minute: steve: text is impossible
   plinss: anyway, we need to wind this up
   plinss: I'm not hearing consensus on moving to 90minute telecons
   dbaron: I would rather not
   several: more frequent telecons
   glenn: I would prefer two a week, one being general and one topic-specific
   glenn: rather than one 90 minute call
   Tab: prefer staggering, better to accommodate other timezones
   plinss: not hearing consensus on any one solution, push aside for now
   <sylvaing> fwiw I'm not sure what a general topic is. modularization is
              also specialization.

Administrative: F2F Planning
----------------------------

   plinss: Offer for F2F meeting in September in Zurich, not sure if idea
           was to replace San Diego or what
   vhardy: There's a conference in Zurich, so SVGWG going there to meeting
   Tab: could throw an FXTF day onto it, but not add CSSWG on top of it
   szilles: does that mean we can do less FXTF stuff in August?
   plinss: don't think we should take FXTF day every F2F
   plinss: So keep CSS in SD in August
   plinss: Don't want to add CSS meeting in September?
   jdaggett: We should do 3 days at TPAC
   Sun-Mon-Tues
   jdaggett offers Tokyo next march
   fantasai: also had an offer for Tucson from Molly for next year
   discussion of meeting dates/places
   February: Tokyo or Arizona?
   shane: we can offer sydney as well
   jdaggett: tentatively, February in Tucson, May in Tokyo
   PENCILLEDIN: February in Tucson, May in Tokyo

Style Attributes PR
-------------------

   fantasai: Style attr has 2 passes from IE and FF, go to PR?
   RESOLVED: Take CSS Style Attributes to PR
   ACTION fantasai: publish test results etc.
   <trackbot> Created ACTION-459

Multicol Test Suite
-------------------

   howcome: bunch of tests at Opera, not all submitted but in process
   plinss: Shepherd shows 42 testcases, harness shows 23
   howcome: I encourage people to try their implementations
   howcome shows off some tests
   howcome asks MS to look at the tests
   and Mozilla, and Chrome
   ACTION Tab: run multicol tests in Chrome
   <trackbot> Created ACTION-460

   stearns: how do you run the tests in Chrome? Tests don't have a prefix.
            Do you run grep on them?
   fantasai: will ideally have build system handle that, for now I'd run a regex

   stearns: We have an ad-hoc policy of asking for reviews, maybe getting
            reviews
   stearns: wondering if we can be more organized, exchange reviews
   fantasai suggests QA people exchange reviews amongst themselves
   hober: does Shepherd email people about new tests to review?
   plinss: who should it email?
   krit: could email owner of test
   plinss: could have owner of test, commenters, owner of suite get emailed
   krit: one thing to write tests, also have to review tests
   vhardy: Should we try to accept tests, and when someone runs the test
           against implementation
   vhardy: they'll report errors
   dbaron: I've been pushing for that for awhile
   Florian: It's the passed for the wrong reason that's more annoying
   ...
   fantasai: could approve tests that pass multiple implementations, but
             should track tests that aren't reviewed manually separately,
             so that if someone does want to go through them they know
             which ones to review
   ....
   stearns: should not have them stuck in Awaiting Review
   plinss: Suggest doing what we did with 2.1, build them into the test
           harness, run the tests, shift them into approved once the test
           suite seems stable, but don't give them reviewer links until
           they're individually reviewed
   <howcome> Here's an alternative way to get to the multicol tests
   <howcome> http://test.csswg.org/source/contributors/opera/submitted/multicol/
   ...
   * dbaron wonders what issue we're trying to solve now

Received on Tuesday, 15 May 2012 17:35:44 UTC