[CSSWG] Minutes and Resolutions 2010-12-15


   - Reviewed status of PFWG comment on CSS3 Backgrounds and Borders
   - Argued over whether the 2007 Snapshot should exist, prompted by
     the Director's suggestion that it be a NOTE instead of a REC.
   - CSS2.1 Test Suite RC4 was published over the weekend. Updated
     implementation reports are requested. Testers are encouraged to use
     the WG harness, as this grandfathers results from unchanged tests
     and presents those tests that need retesting.

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


   César Acebal
   David Baron
   Bert Bos
   Arron Eicholz
   Elika J. Etemad
   Simon Fraser
   Daniel Glazman
   Koji Ishii
   John Jansen
   Brad Kemper
   Hĺkon Wium Lie
   Peter Linss
   David Singer
   Daniel Weck
   Steve Zilles

<RRSAgent> logging to http://www.w3.org/2010/12/15-CSS-irc

Scribe: fantasai


   <szilles> Can we add CSS-Beijing-2007 to the agenda
   Peter: Any other agenda items?
   Arron: I have one about tracking Bert's edits
   Sylvain: PFWG?
   Peter: already on the agenda

PFWG comment on css3-background

   glazou: Did my action item to send official response to PFWG.
   glazou: We should hear back from them

Cross-WG Comments

   glazou: A year ago we discussed whethers editors should bring everything
           back to the WG concall
   glazou: or can the editors handle some things themselves
   glazou: Apparently this is an issue for some other working groups.
   glazou: When a comment comes official from a WG, it should go back to
           the conference call.
   glazou: Otherwise we have no means to say the answer of a given member
           is the position of the WG.
   glazou: It's a little change from what we decided awhile ago, but it
           seems necessary.
   Bert: I don't think it means the answer has to come from the chair,
         but it has to be clear that it comes from the WG.
   Steve: Would it suffice ...
   glazou: There's a thread, and lots of responses.
   glazou: Somehow have to designate that one of them is the official
           position of the WG.
   Steve: Sometimes we have a conclusion, but don't have exact wording.
   Steve: Sometimes it's useful to designate an existing response as official.
   glazou: We'll have to decide on a case-by-case basis.
   fantasai: I think if we're dealing with PFWG, because they expect more
             formality, we have to have every single email response be an
             official WG response, whether it's asking a question or
             suggesting an edit or what. But with i18n or SVG, we might
             not need to be quite as formal.
   glazou: We need to have a WG position on each cross-WG issue, in case
           another WG has the same comment.

CSS Beijing / Snapshot 2007

   Peter: Question from directory is whether should be REC track or WG Note
   Steve: It's important to be REC track, because this is how we have
          defined rolling out CSS as a series of modules.
   Steve: We need to establish when we're establishing a new conformance level
   Steve: The doc does define conformance criteria.
   Steve: It doesn't work as a note, because that isn't something you can
          conform to.
   Steve: There are certainly other groups, e.g. SVG and.. Timed Text's group..
          that develop profiles of their specifications as conformance levels.
   Steve: Bert, you were on the phone call. Why did Ralph not think it was
          a REC-track document?
   Bert: It doesn't have any conformance requirements, and it's an
         informative document.
   fantasai: It's not a profile, so much as defining all of CSS as of a
             particular point in time.
   Steve: What defines conformance to CSS?
   Peter: The test suite for that spec.
   fantasai: No, that doesn't define conformance, it helps measure it.
   glazou: Does anyone care about this set of specs as a conformance target?
   glazou: Would anyone say "We implement Snapshot 2007"?
   Sylvain: I wouldn't.
   Peter: The document reads like a Note.
   <dsinger> I guess I can see some point to being able to say "We implement
             CSS3 2010" (and have 2011 include more modules)...
   Steve: My question is, should there be a REC-track document, and how
          should that be written.
   Steve: Some people are making tools for this.
   Steve: Such people want to know what is going to be the next level for
   Steve: What is the next 2.1
   Steve: Tool vendors need to be able to expect what comes next.
   Steve: Notes are not as authoritive

   fantasai: There are a couple things that are normative and should be
             captured normatively somewhere.
   fantasai: The first thing, it defines in what order the specs modify
             each other.
   fantasai: Since that was an issue someone raised.
   fantasai: The ordering in section 3 is normative
   fantasai: Second, it defines what to do with a partial implementation
   fantasai: how to ignore values and things like that
   fantasai: The the third thing is, it gives recommendations for prefixing --
             when your'e supposed to prefix, when you're not supposed to prefix.
   fantasai: Those are the three things that are normative in this spec,
             and are not captured elsewhere.
   fantasai: We can split the spec into a normative spec and an informative note
   fantasai: But I don't think we should drop these on the floor because
             the document as a whole reads as a note.

   Sylvain: Who uses this?
   fantasai: Validator people use it to guide their implementation
   fantasai: Anyone trying to figure out what the state of all these CSS3
             modules is would find this useful.
   fantasai: It would replace the CSS3 Roadmap, which presented a different
             view of modularization.
   glazou: Does anyone in our group use this document?
   Steve: We are not the target audience. It's the people outside our group.
   dbaron: *We* all know what it says.

   fantasai: This is the replacement for CSS2.1
   dsinger: A lot of people talk about CSS3
   dsinger: This document would define CSS3
   <dsinger> would it not be most helpful to publish snapshots that include
             only modules that are 'done'?
   Sylvain: Nobody cares about 2007 Snapshot
   Steve: Do you see value in the 2010 Snapshot?
   Sylvain: I don't see the value of it.
   Steve: It's the replacement of 2.1 for what the next "version" of CSS is.
   Steve: It's attempting to define a set of things that are intended as
          where the CSS group sees the next collection of things coming together.
   <bradk> who is linking to the snapshot: http://www.google.com/search?q=linkto:%20http://www.w3.org/TR/css-beijing/
   fantasai: The state of our modules is a mess. Nobody knows what is
              stable from its status.
   Sylvain: Then that's the problem we should solve
   Steve: Changing the status tracking method is not giving us a target
   * fantasai could somebody take over minutes so I can join PFWG call?
   ScribeNick: dbaron
   * fantasai thanks
   sylvaing: Just make it a note and move on with the 2010 snapshot.
   SteveZ: OK with me.  I agree with the comment that says we shouldn't
           use that name, but...
   <dsinger> defining snapshots that include stable modules that are
             implemented ('widely') to give people a meaning behind
             'CSS3 2010' makes sense
   sylvain: I don't see why the WG should be arguing about taking a
            snapshot from 2007 to CR.
   peterl: There is also pushback that this is a 2007 document that
           should be immediately replaced by a 2010 document.
   <dsinger> defining snapshots that include modules that might be done
             in the future or modules that, though done, are not (yet)
             'mainstream' makes much less sense
   peterl: As a note, it can just be published on its own and we don't
           have to wait 3 years.
   peterl: There's also a mistaken impression of css3, which is a
            meaningless term.  We should be defining what is CSS today.
            This can be a note.
   <glazou> +1
   peterl: I think the normative parts of this document can be folded
           into a different rec-track document.
   SteveZ: which rec-track document?
   peterl: CSS 2.1 or a CSS core module
   SteveZ: That's what, to me, the snapshots were to do.
   peterl: I'm not talking about putting the list of modules in 2.1; that
           should just be a note.
   SteveZ: There are conformance requirements in there; the order of that
           list is important.
   glazou: Steve, the conformance requirements for css3 are going to change
           over time?
   <dsinger> there is no 'css3'
   <dsinger> there is 'css3 2010', 'css3 2012'
   dbaron: We're not doing a single CSS3.
   glazou: That's something we can understand, not users.
   <dsinger> the alternative is css3.0, css3.1, and so on...
   dbaron: The point of this document was to explain that.
   glazou: Given that this document is unknown in the designer community,
           I think it's a failure.
   sylvaing: "css3" is a mess; any time anyone submits a draft, it's
             css3-something.  We could make it css-something until we agree
             it's part of css3.
   sylvaing: Nobody cares about a document that describes the state of the
             world in 2007.
   SteveZ: It's also the state in 2009.
   SteveZ: I'm not arguing that we should do 2007 and not 2010.  I really
           don't care which is the first doc that comes out.  We do need a
           document that says "this is what css3 is today and this is what
           conformance to css3 would mean today".
   SteveZ: And this document, once published, doesn't change.
   sylvaing: Do we as a WG need to spend another half hour moving to CR the
             2007 version of that snapshot?
   SteveZ: No, as long as we do it for 2010?
   peterl: I think there's a valid question whether the 2010 snapshot should
           be a rec-track document.
   SteveZ: Where would you put the order of conformance?
   arron: css3-mediaqueries depends on CSS21, so its conformance requires
          support for the others
   dbaron: I think part of the issue with the ordering is that we need to
           say what overrides things in other specs, not just what is required.
   glazou: I don't think we need a REC.  I think an unofficial Web page
           from this WG is enough.
   SteveZ: I don't think it's good enough for claiming conformance.  And
           there's an issue of other groups defining conformance to CSS.
   glazou: css3 itself is not a spec.  So conformance to css3 means nothing.
   peterl: If this is a REC-track doc, where is its test suite?
   SteveZ: Test suites of individual modules.
   dsinger: Seems all it needs to say is "you must be conformant to the
            following modules"
   SteveZ: And that's what it says.
   peterl: Do we need to take that through the REC track?
   SteveZ: yes

   peterl: CSS is a moving target until this WG stops publishing docs
   SteveZ: not acceptable
   dsinger: And the point of the snapshots is that CSS 2010 is a stable target.
   peterl: Can't that just be a note?
   SteveZ: Not if you're going to define conformance.
   SteveZ: ... I think they have to be targets for a given market.  The
           snapshot might define multiple sets for different markets.
   SteveZ: Without that, you have no guarantee that different products will
           behave the same way.
   SteveZ: And we're back in the "good old days" of competing implementations.
   SteveZ: And part of the process of what goes in the snapshot is that
           there's agreement that the people who are doing this thing to be
           working towards.
   SteveZ: Most implementations don't work very well with changing conformance
           in midstream.
   sylvaing: Which company today decides what they're going to do in CSS based
             on a snapshot from the WG?
   SteveZ: None that I know of, but I hope that they would.
   dbaron: Snapshot can also be following the implementors rather than leading.
   <glazou> to be clear, I still think this document is useless, I said it
            many times in the past ; I find it a waste of time and energy,
            disturbing us from MUCH more important things on our radar ;
            if you count the fact that just nobody ever refers to it, I think
            we should dump it for good
   SteveZ: A tool provider can make a statement that this tool produces code
           that works with css3 part 1; it's much simpler than a list of
           products and versions.
   sylvaing: Does Adobe refer to snapshots in documentation of their products?
   SteveZ: We don't trust CSS3.
   peterl: This is a three year old document, and it had to be because of the
           levels of the modules it's referring to.  And if it's a REC-track
            document, we'll always have that problem.
   SteveZ: That's fine.
   <glazou> szilles: can you please explain "don't trust" ?
   peterl: And that means we'll be publishing snapshots out of date.
   dsinger: But that's exactly right.
   dsinger: You can always implement the modules ahead of the snapshot.

   sylvaing: Do we need a snapshot document to do this?  Every time somebody
             writes a module it gets tagged "css3".  Maybe we could not add
             "css3" until it's done?
   dsinger: I think putting "css3" in the module names is confusing.
   dbaron: I think we did agree to drop that about 4 years ago, but we never
   glazou: Even if we drop it, Web authors will still use it.
   glazou: For them, it's the next version of CSS, called CSS3, whether you
           like it or not.
   SteveZ: But they can't go look at the set of specs called "css3".
   dsinger: The rest of the world is talking about CSS3, and I think we need
            to give that a definition.
   glazou: We have one document, "Selectors", without CSS3 in the title, and
           everyone calls it CSS3 selectors.
   dsinger: Someone should look through the module names and come up with
            recommendations for changes, and whether we should do this bundling.
   <Bert> (Selectors does have level 3, namespaces doesn't)

   SteveZ: You'll get lack of agreement about which of the modules constitutes
           a reasonable set for interoperability.  I don't care whether it's
           documenting semi-future or the past.  I think it's useful to have a
           set of things that are frozen in time, and you can have more than
           one of these sets (over time).
   SteveZ: What I find surprising is that I thought we had this discussion
           when we created the snapshots, but we seem to be bringing up the
           same things that caused us to create snapshots in the first place.
   sylvaing: Can we agree that ... ?  I don't see the utility of debating
             moving 2007 to CR.
   SteveZ: I don't care about 2007, but I care about 2010.
   SteveZ: I believe the text for 2010 is going to be basically the same as
           the text for 2007, except the module list.
   <glazou> sorry guys, cannot rejoin the call
   peterl: Out of time, not sure I'm hearing consensus.
   SteveZ: I don't think we have consensus yet.
   SteveZ: I think a number of us believe a snapshot that documents a stake
           in the ground is useful, and another group believe that ...
   glazou, it's past the end time

   SteveZ: We're ducking the issue of that we need to work at publicizing it.
   SteveZ: So if we do this we need to make sure people understand what it is
           and how to use it.
   sylvaing: I don't think it has to be a REC.  But it needs to be useful and known.
   sylvaing: If nobody finds it there's no point.
   SteveZ: I think the intent was it would show up under /TR/CSS3
   <dsinger> I guess "CSS3.2 is defined as containing the following modules"
             is OK, not as good as a conformance statement
   peterl: /TR/CSS
   peterl: Suggestions on how to move forward?
   sylvaing: Not move 2007 through the whole transition request until we
             figure this out?
   dsinger: ... and remove "css3" from names of modules.
   * glazou hopes you all understand that releasing a CSS 3.X every time we
            have a new stable module is crazy from a user's perspective
   peterl: I believe we have agreement there.  Question is whether /TR/CSS
           needs to be REC-track or can be a NOTE.
   peterl: Maybe discuss over email?
   dsinger: Also ask people at W3C who objected to it being a REC-track
   glazou: Ralph?

Test Suite

   peterl: We have RC4 online, and now need more data
   peterl: Harness will go to the tests we need the data the most for.
   peterl: We need more results to get good blocking data.
   peterl: So work on that for next week?
   peterl: There's a teleconference next week but not the week after.

Received on Wednesday, 15 December 2010 23:15:50 UTC