[CSSWG] Minutes Telecon 2012-09-26

Summary:
   - RESOLVED: Add Brian Birtles (Mozilla) as editor of CSS Masking
   - RESOLVED: Minimum size for running reftests is 600x600;
               tests should be written to fit within that size.
               Recommend that UAs use a rectangular viewport that is larger
               than that if possible, to avoid missing errors due to symmetry
               of square viewports.
   - RESOLVED: FPWD CSS Counter Styles Level 3
   - Discussed interpretation of IRIs in url() notation
   - Discussed shrink-to-fit sizing of multi-column elements
   - Discussed relationship of line grid module and line layout module
   - Noted specs missing from priority lists
   - Plan to take CSS3 Conditional to LC on October 10th, please review
     and file any issues before then.

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

Present:
   César Acebal
   Rossen Atanassov
   Tab Atkins
   Bert Bos
   Tantek Çelik (via IRC)
   Arron Eicholz
   Elika Etemad
   Daniel Glazman
   Rebecca Hauck
   Molly Holzschlag (via IRC)
   Koji Ishii
   John Jansen
   Brad Kemper
   Peter Linss
   Edward O'Connor
   Anton Prowse
   Florian Rivoal
   Simon Sapin
   Dirc Schulze
   Alan Stearns
   Leif Arne Storset
   Lea Verou

<RRSAgent> Logging to http://www.w3.org/2012/09/26-css-irc
Scribe: fantasai

Administrative
--------------

   glazou: any extra items?
   krit: Would like editor to Masking
   krit: Would like to add Brian Birtles from Mozilla
   krit: He's already worked on SVG masking spec
   glazou: Any objections?
   RESOLVED: Brian Birtles added as editor to CSS3 Masking

Counter Styles
--------------

   <glazou> http://lists.w3.org/Archives/Public/www-style/2012Sep/0464.html
   glazou: Request from fantasai and Tab to publish FPWD of counter-styles
   Florian: This is split out for CSS3 Lists?
   fantasai: yes
   Florian: It's removed all the stuff except CSS2.1 / CSS2.0 stuff?
   fantasai: Yes, except ethiopic-numeric, because it can't be represented
             by @counter-style. This is marked as an issue.
   Florian: What about katakana / hiragana?
   Florian: There was some question of how they behave, why not look at the
            implementations that exist?
   fantasai: No disagreement there
   Florian: If there's no implementation, then remove them.
   glazou: We have enough time to remove them in the future
   glazou: Noticed one of the latest edits was edit to type attribute in
           CSSOM (changed to system)
   glazou: There was some discussion about that
   glazou: Change seems good to me.
   Bert: Can we change the shortname? Seems odd for CSS3 shortnames to be
         inconsistent
   Florian: Didn't we resolve to change it to the new pattern?
   <TabAtkins> Yes, it's the new pattern that we'll change everything to.
   Florian: Eventually we will migrate everything to the new scheme
   glazou: We resolved on that last week
   Bert: Think there's better things to do, but ok.
   glazou: We need to. The current naming scheme has been inconsistent,
           and it's confusing people.
   Florian: I will object if they are not marked at-risk
   fantasai: They're marked
   glazou: It's in the status section
   Florian: Ok, I'm good.
   RESOLVED: FPWD CSS Counter Styles Level 3

   * glazou reminder to all : read the conf call minutes as soon as they
     are released and object - if you have to - as soon as possible

Viewport Size for Reftests
--------------------------

   <glazou> http://lists.w3.org/Archives/Public/public-css-testsuite/2012Sep/0020.html
   leif: Deferred so I could investigate what Opera needed.
   leif: Essentially have no problem with 600x600
   leif: Might need different size in the future, but probably would
         prefer to mark that with metadata
   fantasai: If we need a smaller size, then we should figure that out now,
             not go back and mark tests that fit within the smaller size
             and say the others can't be run on that size
   leif: ... viewport tests probably need a special keywords ...
   arronei: We should add a flag for tests that don't fit within 600x600

   rossen: From past experience, when we had tests that rely on a square
           container, sometimes this can hide some buggy implementations,
           especially when dealing with different permutations of directions
           / writing modes / etc
   rossen: when both containing width and height are the same, might hide
           a buggy implementation there
   <antonp> +1
   rossen: if you're settled on 600x600, that's fine, but a square container
           is prone to hiding some bugs
   krit: Mozilla uses 1000x1000, and Webkit uses 800x600
   rossen: 1000x1000 doesn't solve the issue I was talking about, but
           800x600 would be perfect
   <SimonSapin> +1, I had such bugs. (Non-square viewport is better)
   rossen: also @media aspect ratios, setting square viewport can hide
           some things that you want to find out sooner than later
   fantasai: This isn't the size to use, it's the smallest size that you
             can make the viewport and still capture all relevant data in
             a screenshot

   glazou: Enough info to make a decision?
   fantasai: Yeah. We'll go with 600x600, and recommend that UAs use a
             rectangular size that is larger than that if possible.
   Florian: I'm mildly surprised no one needs a smaller size for mobile
            testing
   leif: Often the desktop viewport size is used
   leif: If we wanted to go with the smallest mobile screen size, we'd be
         at 240.
   leif: That's a bit small.
   RESOLVED: as fantasai summarized above
   <florian> I am wondering if the size limit implies constrains on the
             fonts that you can use when running the tests, as some fonts
             may cause the text to overflow

Multi-column Shrink-to-fit
--------------------------

   <glazou> http://lists.w3.org/Archives/Public/www-style/2012Jul/0598.html
   Sapin: The part of multicol about shrink-to-fit doesn't make sense.
          We should just remove it.
   fantasai: I agree with removing any implication that css3-multicol
             defines multicol intrinsic sizing
   Sapin: We can define it later in css3-sizing if necessary.

   Bert: I didn't understand what the problem is
   Sapin: css3-multicol sizing uses shrink-to-fit sizing when available
          size is none, but CSS2.1 [...]

   glazou: You said there is little interop in your email
   Sapin: If we we make float multicol elements, we can see intrinsic size
          without defining width
   Sapin: Results are all over the place
   fantasai: I raised an issue about multicol intrinsic sizing that Hĺkon
             thought was in the spec not making sense anyway
   glazou: Impact on the spec?
   Sapin: [...]
   anton: Make it explicitly undefined?
   Sapin: Yes, we can add that it's undefined
   Sapin: I don't know if, for example, Flexbox explicitly says the
          preferred width is undefined
   fantasai: No, we define it.
   Florian: Easier for us and implementers if we make it explicitly undefined

   rossen: Have another question here, can you elaborate what you meant by
           "results are all over the place?"
   11:30 -!- Rossen [Rossen@131.107.0.115] has joined #css
   Sapin: For example, some browsers takes the preferred width as if the
          element were not multi-column, and just use that
   Sapin: Some browsers multiply the results by the column-count
   Rossen: Were there any browsers doing what the spec is asking for?
   Sapin: I don't know what the spec is supposed to be asking for
   fantasai: I think we should make it's undefined, because the desired
             behavior is unclear/unknown, and we don't have interop

   Bert: It's already undefined in the spec, what did you want to change?
   Sapin: I don't know how having an unknown available-width is useful
   glazou: Seems to me some people need to do more investigation to resolve
           this
   rossen: We did a bit of work on this, I need to test a little on that
   rossen: We read the spec at the time, and it kinda made sense
   rossen: I would prefer if we can take this next week and then have a
           few days to look around and see what exactly would that mean
   glazou: We'll return next week

   anton: Ca we get confirmation exactly what the proposal is
   Sapin: Remove lines 3-10 inclusive
   Florian: Seems we have to remove something, but unsure what the scope
            of that its

URL notation and IRIs
---------------------

   <glazou> http://www.w3.org/mid/4FBB2B56.9080805@mcc.id.au
   <SimonSapin> URLs are ASCII only (and even more restrictive than that)
   TabAtkins: We just have to make sure our definition of URLs includes IRIs
   glazou: How many changes do we need?
   TabAtkins: Just in css3-values
   Florian: Some people complained that we shouldn't use the term URL when
            we mean IRI
   TabAtkins proposes ignoring that request.
   TabAtkins: Everyone calls them URLs.
   krit: CSS3 Images need to update to CSS3 Values and Units instead of
         CSS2.1 then
   glazou: Editing CR?
   fantasai: This qualifies as a clarification, since this is what was meant
   fantasai: I think I even had an issue marked on what the correct terminology
             should be to include IRIs, and nobody ever commented on it
   fantasai: So I just removed the issue.
   <SimonSapin> RFC 3987 instead of RFC 3986
   glazou: Also need to update the prose of the spec

Line Grid proposal
------------------

   <glazou> https://lists.w3.org/Archives/Member/w3c-css-wg/2012JulSep/0372.html
   Florian: At the Kyoto F2F, fantasai made a line grid proposal
   Florian: Everybody liked it, and since there was the Line Layout module,
            people wanted to put Line Grid into Line Layout
   Florian: But since then it seems Line layout  is complicated and moves
            slowly
   Florian: Propose making it its own module, and moving it forward
   Szilles: Line grid is equally complicated, because you don't know what
            to align with the line grid
   Florian: I agree there are ambiguities in what will happen until Line
            Layout is finished
   florian: But [...]
   szilles: No, it's because of the way line-height is defined
   szilles: You don't know where the baseline is within a line
   fantasai: Isn't that only true when something is vertical-align: bottom
             or top ?
   szilles: don't recall, but not about differing baselines
   Florian: Seems to me in generalized situation what you're saying is right,
            but there are a lot of cases that would work
   szilles: I guess I disagree, since if you don't figure this out from
            the beginning you won't get it right
   szilles: But I will commit to having a Line Module layout by Lyon
   Florian: Couldn't we still have different modules that depend on each
            other? Or is dependency too strong?
   szilles: you need a definition of lines that works
   anton: I think you'll end up relying on dependencies that aren't thought
          through
   glazou: Since Steve's committed to giving us a module by Lyon, I suggest
           waiting until then
   Florian: Ok, that's fine

More Administrative
-------------------

   glazou: end of agenda, anything else to discuss?
   fantasai: Do we have dates for F2F in February yet?
   molly?
   glazou: Proposed 4-6 of February in Tucson
   Florian: pending confirmation
   glazou: I have an exclusion in February, there's a W3C workshop about
           EPUB in NYC 15-16 of February
   glazou: So would like to avoid those dates if we ever change
   szilles: don't think molly can do those dates anyway
   fantasai: So just need Molly to confirm dates?
   <molly> It is dependent upon availability and sponsors - but I will
           leave those days out

   glazou: Anything else?
   Florian: Yes, about your priority lists, seems to me there were some
            specs missing. Maybe re-issue an updated list?
   glazou: People already commented that they were missing, so I hope
           responses included them
   fantasai: If they didn't include them, need to go back and ask
   Florian: Public responses missed e.g. CSS Variables
   glazou: Ok, I'll deal with that
   glazou: Peter and I will go through responses before TPAC so we can
           discuss, so really need responses soon so we can aggregate
           data
   * molly needs dates of next SVG meet too is it in feb?

@supports
---------

   fantasai: For CSS3 Conditional, Tab and I plan to request LC on
             October 10th telecon, that's in 2 weeks. Please review
             and send issues before then.

Meeting closed.

Received on Thursday, 27 September 2012 23:05:25 UTC