[CSSWG] Minutes and Resolutions 2008-07-16

Summary:

   - Discussed charter. CR exit criteria has errors, and everything's dependency
     on CSS2.1 needs to be noted.
   - Discussed Color module. Publication request will be sent out today.
   - Discussed new test suite harness. See
       http://lists.w3.org/Archives/Public/public-css-testsuite/2008Apr/0027.html
     for background information. (The prototype is currently running as Member-only
     until it's been vetted for correctness/security. It will eventually be public
     and open source.)
   - Discussed need to build a test management system to handle submissions,
     reviews, and test suite coverage reports.
   - Discussed whether forming a CSS Test Suite Interest Group would be a useful
     exercise, if it would generate interest and activity from people not in the
     WG or only overhead for W3C.

===============================================================================

Attendees:

   David Baron
   Bert Bos
   Giorgi Chavchanidze
   Arron Eicholz
   Elika Etemad
   Ming Gao
   Melinda Grant
   Anne van Kesteren (via IRC)
   Peter Linss
   Alex Mogilevsky
   David Singer
   Jason Cranford Teague

ScribeNick: fantasai
<anne> Regrets + anne
<anne> (also for the next three weeks, as I'm on holiday)

Charter
-------

   Peter: I updated charter with Chris's feedback
   Peter: Any comments?
   <plinss> http://www.w3.org/Style/Group/2008/draft-charter2.html
   <plinss> http://www.w3.org/Style/Group/2008/proposed-charter.html
   Melinda: I have a concern
   Melinda: Bullet number three we say "..."
   Melinda: We should say "for each feature"
   Melinda: It sounds like we musth have two complete implementations of the
            entire CSS2.1 rather than two implementations of each feature, etc.
   Peter needs to check the process document
   <melinda> http://www.w3.org/2005/10/Process-20051014/
   <melinda> http://www.w3.org/2005/10/Process-20051014/tr.html#cfr #2

   David: Why was the wording about potentially merging and splitting modules
          taken out?
   Peter: Since adding new modules would be done by amending the charter anyway

   David: What would happen with the SVG features module that's being discussed?
   David: Does that require rechartering?
   David: Does it require amending the charter?
   David: Or can we just do it?
   Melinda: I didn't see any wording in the charter said that items not in the
            deliverables list can't be advanced along REC track
   Peter: That's how we plan to interpret it, but I didn't want to specifically
          call that out
   Peter: We do want to focus on our deliverables and not get sidetracked
   Peter: but I also don't want us to get stuck in charter process
   Peter: Wrt CSS-SVG thing.. I'm not sure if that fits our definition of
          working with other groups
   Melinda: We might want to ask Chris how he thinks we should handle these kinds
            of emergent things
   dsinger: On the subject on other material, we can always discuss things on
            the mailing list and work on it
   dsinger: but getting formal time on it, that requires having it in the charter
   Peter: our focus should be driving our deliverables to REC
   dbaron: I can see this CSS-SVG thing advancing quickly enough that we might
           have multiple implementations by the end of this charter period
   * Bert notices a symmetry: after 2 Recs in 11 years, we now promise 11 Recs
          in 2 years...
   dbaron: Do we want to just let that slip?
   Peter: how do we split this with SVG?
   dbaron: I think if we're adding features to CSS, it should be in CSS
   Peter: Isn't there something about changing interpretations of SVG?
   dbaron: there might have been a few details
   <melinda> Bert, what could go wrong?
   <Bert> :-)
   Peter: I'll make a note to ask Chris about cross-group issues, where that
          should live in our charter, using this as an example
   ACTION: Peter talk with Chris about cross-group issues

   Bert: I think 11 RECs is a bit much
   Bert: We have a list of priority things...
   Bert: You really think we can make 11 RECs in just two years?
   dbaron: Why did Paged Media get bumped up?
   Elika: We're planning to publish Last Call this year
   Elika: Doesn't make sense to amend charter right after it gets approved
   Peter: A lot of the things in the list are small and/or far along
   Elika: A lot of them depend on CSS2.1
   Melinda: I think even if we only worked on CSS2.1, it would be a stretch
            to finish it in 2 years
   fantasai agrees
   Bert still thinks variables should not be on our charter
   Bert: That's an architectural concern, not a process concern

   Peter: I hear your argument, but if we have implementors who want to work
          on it, we need

   <dsinger> perhaps each of the 11 should have a calendar, to show what
             needs done by when
   dsinger: I wonder if for each of the 11 we have a calendar to show what
            needs to be done by when
   dsinger: otherwise we'll procrastinate until the end of the 2 years and
            realize that we can't finish them all
   Melinda: We maybe don't need it in the charter, but it would be a good
            companion document

   Peter: Do we call out dependencies anywhere?
   fantasai: Most docs depend on CSS2.1. If it doesn't make it, most things
             won't make it
   Melinda: only Selectors doesn't
   fantasai: Should put that in the charter. Any other dependencies can be
             tweaked out, but anything that depends on CSS2.1 depends on CSS2.1
   fantasai: If 2.1 doesn't make it to REC, almost nothing else will no matter
             how ready
   fantasai: Only Selectors and Media Queries are independent
   Peter: I will call out the dependency on 2.1 in the charter
   ACTION: Peter update charter in response to Melinda's comment on CR crit
           and 2.1 dep

   <anne> (FWIW, I think both Media Queries and Selectors have a grammar dependency on CSS 2.1)

Color Module
------------

   Peter: We have permission to publish LC
   Peter: Chris says we don't need to ask for permission here
   dbaron: That was the transition request. What about the pub request?
   dbaron: Did you request a publication date?
   Peter: no
   dbaron: I'll work on that then
   Peter: Where are we with implementation reports?
   dbaron: We're in good shape for implementations, but don't have reports
   Peter: Can we generate those by the end of the LC period?
   dbaron: Once the LC is published, the test suite will reflect a published spec
   dbaron: then we can request implementation reports
   ACTION: dbaron prepare implementation report template for CSS3 Color

Marquee
-------

   Peter: Ready for LC?
   fantasai: no. Still some issues with marquee-direction table
   fantasai: sent message to www-style this morning
   ...
   Peter: Let's not get into the technical discussion here
   Peter: revisit in a week

Test Suite Harness
------------------

   Peter: Wanted to let everyone know that there's a demo version online
   dbaron: what does it do?
   fantasai: Records pass/fail results. Anyone can stop by, run a few tests,
             and leave. It can compile implementation reports from that. Uses
             UA string to identify an implementation.
   <Bert> http://www.w3.org/2008/07/test-harness-css/
   Peter: the goal is to make it easy to do implementation reports
   <fantasai> MWI test harness it is based off of :
              http://www.w3.org/2007/03/mth/harness
   fantasai: it's Member-only until Dom has a chance to review the code
   * Bert sees the report generator currently crashes :-)

   Peter: The other thing I want to discuss, we've been tossing idea of
          building a test management system
   Peter: Allow people to submit tests, manage reviews and approvals, etc.
   Peter: I don't think there's anything out there, we'd have to build one
   Peter: HP is interested in contributing resources, wanted to put out a
          call to see if anyone else is interested
   Peter: I'm not asking for answers or commitments, just give a thought
   Peter: If there are questions about why or what's the value, let's hear them
   Peter: I think having a system like this rapidly in place would be a big
          win for us, for CSS2.1 test suite in particular
   Peter: If there's something open source out there that we can use, can be
          modified, etc. that will get us there rapidly, let us know?
   Peter: We could use any able-bodied hands that can write PHP or whatever

   dbaron: I wonder if we're being too picky about review reqs for the test
           suite
   dbaron: Maybe we don't need as formal a review process as we have
   dbaron: We should be trying to just get tests in
   fantasai: You pointed out that since tests drive interoperability, wrong
             tests drive interoperability on wrong behavior.
   dbaron: Implementors can catch incorrect tests
   Melinda: We *are* catching incorrect tests during the review process

   Melinda: If we collected the thousands of tests on the Web, we could have
            a test suite with a lot of tests. Won't know how correct it is,
            or how much coverage
   dbaron: My worry is that, if I want to contribute tests I don't know if
           the tests I want to write are in progress or if they're actually
           missing
   Peter: That's one problem we want to solve with this test management system
   Peter: It will include tests that have been submitted, tests that are in
          the system
   Peter: One problem is tests ar currently hosted on submitters site, etc.
   <anne> (I agree with dbaron that implementors will easily catch incorrect
          tests.)
   Peter: I think it would be very beneficial for us to build software here
   <fantasai> (anne, but not incorrect specs, which is something else I've
              been catching)
   <anne> (Implementors are usually the ones catching spec bugs in my experience.)

Test Suite Interest Group
-------------------------

   Peter: any thoughts?
   Melinda: I think it'd just be more overhead unless we identify the set of
            people who can really focus on this
   Bert: I'm sure there are people who are good in making test suites.
   Bert: We are not that kind of people
   Bert: Maybe we aren't attracting that kind of people because we aren't that
         kind of people
   dbaron: I don't think we're not that kind of people
   dbaron: I think the not finishing the test suite is another problem, but I
           don't think you'll be able to pull in random people who aren't good
           CSS people and have them write good CSS tests
   Peter: I think the question is, there are people out there who could be
          involved, would we be more likely to get people involved by forming
          a separate interest group, or is just informally coordinating through
          our group enough?
   Bert: How would that affect the organizations that we represent? Would it
         make a difference to any colleagues?
   Melinda: HP has maybe one person
   Peter: I think HP has been demonstrating that we're dedicated to the test
          suites whether or not there's an interest group
   Jason: I have a request from someone at AOL who is interested in joining
          this group. He might be someone to work on tests
   Peter: I think the benefit of Interest Group is to allow non-W3C-Members
          to be formally involved. I think that's the only thing we'd gain by
          forming an interest group
   Peter: Is that worth the extra overhead? (for W3C, for participants in this
          group)
   Peter: I see advantages and disadvantages
   Melinda: Could ask www-style if anyone would be interested in joining to work
            on tests one day per week
   <dsinger> if making progress on items people care about have dates for test
             suites % completion, I bet we'll see more activity
   fantasai: I think having a test day is a great idea
   fantasai: Mozilla does something like that with bug days
   Bert: I don't think test suites are inspiring enough
   <dsinger> test suites are inspiring if their absence has negative consequences
             (like, you get dropped from the charter and you won't get published)
   fantasai: we have several volunteers on the public test list who are writing
             tests because they think it's interesting
   fantasai: I need help reviewing their tests
   * Bert didn't know that sense of the word "inspiring" :-)
   <dbaron> fantasai, is there a list somewhere of the tests that have been
            contributed that need review?
   Melinda: I'd like to see a milestone schedule for CSS2.1, although i don't
            know how to make one though
   <fantasai> dbaron, I can't remember atm, I'll ping you after the meeting
   melinda: Maybe Elika and I can discuss and toss something out next week
   <dbaron> fantasai, not just for me... the list should be publicly available
            somewhere from Style/CSS/Test/
   <fantasai> yes, you're absolutely right

Meeting closed

<fantasai> I think I had a place to put that, but I don't remember if I
            updated it
* fantasai adds that to to-do list
<fantasai> anne, Implementors may be catching spec bugs, but I'm also running
            into spec bugs just by reviewing tests and noticing that the spec
            doesn't justify their assertions

Received on Wednesday, 16 July 2008 19:05:22 UTC