[CSSWG] Minutes Telecon 2013-02-20

Animations
----------

   - RESOLVED: Add Rossen as editor to Animations
   - RESOLVED: Øyvind's clarification accepted for
               http://lists.w3.org/Archives/Public/www-style/2011Sep/0392.html
   - Discussed reversing timing functions. See follow-up thread at
       http://lists.w3.org/Archives/Public/www-style/2013Feb/0524.html
   - Discussed trigger for starting animations. It's not onload, but unclear what it is.
   - RESOLVED: keyframe rules cascade
   - RESOLVED: Keep pseudoElement on animation events. Mark at-risk. Revisit
               in a few months if it's a web-compat problem.

Publications and Stage Transitions
----------------------------------

   - RESOLVED: New WD for CSS3 Paged Media
   - RESOLVED: Allow pseudo-class combinations for @page selectors
   - RESOLVED: Publish css-print with fantasai as editor, updated changes section.
   - RESOLVED: css3-conditional to CR
   - RESOLVED: New WD counter-styles, expect LC in 2 weeks or so

Miscellaneous
-------------

   - RESOLVED: Add percentages to column-gap, not to column-width (use
               column-count for that, it's better)
   - RESOLVED: Push font load events out to separate spec
   - Goodbye to Sylvain! Leaving role as MSFT CSSWG rep to go to Adobe...

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

Present:
   Tab Atkins
   David Baron
   Bert Bos (via IRC)
   Tantek Çelik
   John Daggett
   Elika Etemad
   Simon Fraser
   Sylvain Galineau
   Daniel Glazman
   Rebecca Hauck
   Dael Jackson (observer)
   Dean Jackson
   John Jansen
   Brad Kemper
   Peter Linss
   Alexis Menard
   Ted O'Connor
   Anton Prowse
   Florian Rivoal (late)
   Simon Sapin
   Dirk Schulze
   Nick Van den Bleeken
   Steve Zilles

<RRSAgent> logging to http://www.w3.org/2013/02/20-css-irc
Agenda: http://lists.w3.org/Archives/Public/www-style/2013Feb/0461.html
Scribe: fantasai

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

   glazou: Extra items?
   krit: Edited Masking spec, would like to ask for review by email, is that ok?

   SimonSapin: Proposal for adding percentages to column-width/column-gap
   glazou: You wondered if doable in CR. Let's discuss that after animations

   glazou: Anything else?

   * dino agenda += moment of silence for sylvain

   glazou: Start with Animations, b/c sylvain on call for last time

Animations
----------

   -- Editorship --
   sylvaing: Any objections to adding Rossen as editor to Animations?
   RESOLVED: Add Rossen as editor to Animations

   -- Case-sensitivity --
   <glazou> LOL case sensitivity...
   sylvaing: Clarification for me, case-sensitivity of user-defined idents
             was resolved in Tucson. Is that in CSS3 Values yet?
   TabAtkins: Don't think edit has made it in yet, but will do today.
   sylvaing: Ok, will refer to that.

   -- Multiple values --
   <sylvaing> http://lists.w3.org/Archives/Public/www-style/2011Sep/0392.html
   sylvaing: Current wording ignores multiple values, multiple keyframes, etc.
   sylvaing: Øyvind proposed some text.
   [see email]
   sylvaing: I think good idea to clarify that.
   <dino> +1 to clarifying that text
   dbaron: There was an interop issue with definition of valid @keyframes rule
   dbaron: Previous decision on that, make sure it's clear too
   RESOLVED: Øyvind's clarification accepted for
             http://lists.w3.org/Archives/Public/www-style/2011Sep/0392.html

   -- Reversing Timing Functions --
   <sylvaing> https://www.w3.org/Bugs/Public/show_bug.cgi?id=14805
   sylvaing: Question from dbaron on using timing functions inside keyframe
             rules
   sylvaing: Understood that if you have timing function on keyframe rule,
             ... next one
   sylvaing: What if you're going in a different direction, e.g. reverse /
             alternate?
   <dbaron> http://lists.w3.org/Archives/Public/www-style/2011Mar/0744.html
   sylvaing: When you go from N to N+1, you always use timing function
             defined in frame N.
   dino: Intended, but not written. Sorry. Yes.
   TabAtkins: Timing function on a keyframe defines that particular gap,
              regardless which way you're going.
   dino: Maybe add to spec that animation literally runs in reverse.
   dbaron: Except it doesn't
   dbaron: I don't think it does.
   dino: ?
   dbaron: Do we reduce the math of the timing function, or use the appropriate
           timing function?
   dbaron: e.g. if you have ease-in(), then you ease-in() to that point
           regardless of direction.
   glazou: Right, so we are not reversing animation per se
   [...]
   sylvaing: What does Gecko do?
   dbaron: Have to check
   dino: we process value from 0 - 1, [...] so timing function does actually
         play backwards when you go in a reverse order
   dino: If you're 10% through reverse, we calculate as if 90% of going forwards
   glazou: Ok, let's defer resolution of this issue until dino's email
   sylvaing: Yes, and we should check implementations. I think we do what
             Gecko does
   <dbaron> you mean, you think IE does the same thing that I *think* Gecko
            does? :-)
   sylvaing: But don't think it's a major breaking change if we have to
             swap it.
   smfr: Think we don't what to change WebKit. And intent *is* that reverse
         is a mirror image of the forward animation.

   -- Start of Animation --
   <sylvaing> https://www.w3.org/Bugs/Public/show_bug.cgi?id=15848
   <glazou> http://lists.w3.org/Archives/Public/www-style/2009Dec/0280.html
   sylvaing: Prose in spec about defining start of animation
   <sylvaing>   # The start time of an animation is the latter of two moments:
   <sylvaing>   # the time at which the style is resolved that specifies the
   <sylvaing>   # animation, or the time the document's load event is fired.
   sylvaing: Hard one to test, and not exactly what browsers do, really
   sylvaing: I'm wondering, what is the point of this statement?
   sylvaing: Or is it really, animation applies when the animation-name
             property is resolved?

   smfr: in WebKit we did start until document load, but that's no longer
         true.
   dbaron: Don't think it was true at the time I implemented animations
           in Gecko
   sylvaing: Also, hard to test. Interesting implementation details, not
             sure how you'd test across browsers.
   sylvaing: Should really say it applies at the time the animation-name
             property is resolved.
   sylvaing: Of course, we're getting to, we don't define when things are
             computed.
   sylvaing: But document load is bogus.

   krit: Does that mean that document is loaded (onload), or all required
         parts loaded (e.g. all style sheets)
   krit: When you have document load, can you be sure style is resolved
         for all parts of the document?
   TabAtkins: [...]
   sylvaing: Why does it matter?
   krit: For SVG, when the document is read for complete rendering, that's
         when animations start.
   smfr: We specced that way for CSS originally, then changed our minds
   krit: Might need to align SVG to this then
   sylvaing: Sadly, don't think we have priority on when things are computed.

   sylvaing: Don't want to leave statement in there that doesn't agree with
             implementations.
   dbaron: I think we need to fix this to match what everyone is doing,
           b/c we all agree that this is wrong.
   sylvaing: Think it goes back to resolution earlier, animation applies
             when animation-name is computed, and have last valid @keyframe
             in sorted order. That's it

   krit: Suppose you have huge document, like HTML5 spec, animation on one
         of first elements that get rendered.
   krit: Do animations wait until whole document is loaded?
   dbaron: we all implemented that they start before the whole document is
           loaded
   krit: Then how do you align animations?

   sylvaing: When this is resolved is up to UA.
   sylvaing: Whenever value computation occurs.
   krit: Can we add a sentence that says this might be more precise in the
         future?
   dbaron: I would prefer to define it more precisely in this spec, even if
           we don't have definitions for all terms

   fantasai: krit's point wrt aligning animations?
   dbaron: We don't align animations.
   krit: That's the point of Web Animations, to align them.
   TabAtkins: Might fix later
   dbaron: Don't think it would be acceptable to fix it later, which is why
           I think we should define it now.
   sylvaing: Later on would be enough content that we would be unable to fix it.
   ACTION: dbaron propose wording for
           http://lists.w3.org/Archives/Public/www-style/2009Dec/0280.html
   <trackbot> Created ACTION-543

   -- Duplicated Keyframes --
   <sylvaing> https://www.w3.org/Bugs/Public/show_bug.cgi?id=21018
   sylvaing: Duplicated keyframes
   sylvaing: If you have N for 50%, you drop previous ones, at least that's
             what spec recommends.
   sylvaing: dbaron suggested at the time maybe we should cascade them.
   sylvaing: Not sure what it means wrt compat, if Gecko does that.
   dbaron: Gecko does cascade them. Has not been a compat problem.
   dbaron: I suspect that if we tried to change it in the other direction,
           might be a compat problem. But this one not so much. Or at least,
           we didn't hit any problems.
   dbaron: I really feel that what the spec says is really just very unlike
           everything CSS does.
   * fantasai agrees
   dbaron: It's the norm in CSS that if you have one declaration block, and
           you have another that has one property, it just overrides that
           one property, not throw out entire previous block. Reasonable
           expectation of authors.

   glazou: The OM for animations only returns one rule for the keyframe,
           not multiple.
   TabAtkins: The OM for keyframes is completely busted.
   glazou: Whatever we decide on this topic, the OM should reflect that too.
   glazou: If we allow multiple keyframes with same key to cascade, then
           findRuleForKey should become findRulesForKey and return multiple
           rules. Otherwise won't be editable.
   sylvaing, TabAtkins: fair point

   sylvaing: I agree with dbaron's point in generally, not very CSS-like
             to have bunch of selector-like constructs, and last one cancels
             previous ones instead of having cascade.
   glazou: I agree

   sylvaing: When you ask for 50% rule, you want all rules that are for 50%,
             you want in order of course. maybe at some point, maybe not
             in this level, give me computed/resolved rule for 50%?
   glazou: Yes, I agree with that, we need that too.
   glazou: You could retrieve that from the current findRule in the OM,
           and if you have multiple keyframes, we need API for that.
   sylvaing: Do we need that for this level?
   TabAtkins: If you're looking for the value for width being animated at
              50%, and specified in different keyframes, if you can get a
              list, then it's easy to iterate the list and get that.
   glazou: You said OM is busted. OM has to be consistent with the prose.
   TabAtkins: We can do a minimum fix, and add to it later.
   glazou: Minimum we could do is remove findRule and add findRules.
   fantasai: You could maybe define findRule to return the cascaded result?
   dbaron: No, it needs to return something you can edit.

   sylvaing: Still need to figure out OM. Are we resolving on cascading
             the keyframes?
   sylvaing pokes dino
   smfr: I think that's fine. Don't think any content has multiple keyframe
         rules, except by mistake.
   sylvaing: OK, so we'll update that. Then have open issue on updating OM
             to give a list of rules.
   sylvaing: and another issue on adding API for combined ruleset
   ACTION: glazou send proposal for updated findRules API for animations
           keyframe rules
   <trackbot> Created ACTION-544

   RESOLVED: keyframe rules cascade
   sylvaing: That's it.

   -- Animation Events on Pseudo-elements --
   dino: Alexis has an issue
   darktears: We do have a problem with the pseudoElement attribute on
              animation events.
   darktears: Someone asked on mailing list, how do I know when animation
              finished on a pseudo-element?
   darktears: You don't know today.
   <darktears> http://lists.w3.org/Archives/Public/www-style/2013Feb/0062.html
   <darktears> http://dev.w3.org/csswg/css3-transitions/#transition-events
   sylvaing: Same problem for Transitions
   darktears: Mozilla people bring issues wrt compatibility
   <sylvaing> transition and animation events expose the same property
   <darktears> right
   TabAtkins: If you just fire plain animation issues with pseudoElement,
              you might get unexpectedly more animation events.
   <darktears> WebKit ships it on Transitions
   dbaron: I think we should try implementing it and see if compatibility
           problem.
   <darktears> it's implemented
   sylvaing: Yeah, there's not a lot of content out there that uses animations
             on pseudo-elements. If only because it was not interoperable.
   sylvaing: Event handler code, wouldn't need to filter for pseudo-elements
   TabAtkins: If it didn't work on WebKit, nobody would have written code
              for it, right.
+Florian
   sylvaing: pseudoElement property on these events is pretty new, so no
             real-world content with event-handling code that checks for it.
   sylvaing: True that more events fired. Could be some breakage, but hard
             to imagine it would be huge.
   darktears: Use cases Boris brought on mailing list were rather exotic
   darktears: Problems and use-cases he saw on real content, but to be very
              honest, was very broken code.
   darktears: Website would be broken if WebKit shipped unprefixed

   sylvaing: ...
   sylvaing: Later add animation on ::before
   sylvaing: Your animation code is not checking for the pseudoElement on
             that element.
   sylvaing: Do your animation event processing too early, there is a risk
             of breakage.
   sylvaing: Not sure what we can do here.
   sylvaing: Strategy of changing event name...
   glazou: If we want the opportunity to change, we can consider that
           real-life use cases are rare enough, still allows us to change.
   glazou: Not necessarily true in near future. So it's right time to do this.
   <darktears> I mean in WebKit we do have it implemented to Transitions
               and will probably ship soon with Chrome. We'll get feedback
               soon
   <darktears> so far nothing showed up

   glazou: Seems we running in circles.
   glazou: Are we proposing to add pseudoElement? Yes/no?
   sylvaing: It's already in there
   glazou: Do we care to remove it?
   TabAtkins: Since objections seem to come from Mozilla, but dbaron's ok
              with trying it, think we keep it.
   dbaron: Will come back with info on this, but takes awhile to ship,
           so in a few months
   <darktears> ok for me
   fantasai: Can mark it at-risk, so won't hold up for CR.
   <darktears> yes
   fantasai: Also it's a "let's try and implement it" change, that's what
             CR is for anyway.
   RESOLVED: Keep pseudoElement on animation events. Mark at-risk. Revisit
             in a few months if it's a web-compat problem.

Paged Media / Print Publications
--------------------------------

   glazou: First one is Paged Media
   SimonSapin: Since we requested new WD, some ppl have started reviewing
               it, so I have some old issues I found that I had lost, and
               some new issues too
   SimonSapin: Some easy to fix, want to fix in next few days. Some I want
               to defer after new WD.
   glazou: Anything really critical that could block WD?
   SimonSapin: Don't think so
   glazou: What do ppl think of releasing new WD of css3-page?
   <sylvaing> is always in favor of new drafts
   RESOLVED: New WD for CSS3 Paged Media

   SimonSapin: We just added new feature, having multiple pseudo-classes
               in @page selectors. New in draft, but we have two
               implementations already.
   <fantasai> e.g. @page :first:left
   glazou: Did you update specificity section?
   SimonSapin: still need to do that, filed an issue
   RESOLVED: Allow pseudo-class combinations for @page selectors

   glazou: Next one, Print Profile.
   fantasai: Needs to be published with Paged Media.
   https://www.w3.org/Style/Group/css-print/
   fantasai: Just switched it to WG Note, and updated references.
   <dbaron> https://www.w3.org/Style/Group/css-print/#section-images
   fantasai: Section on handling image-rendering properties, specifically
             object-fit / object-position.
   fantasai: Previous CSS Print profile included references to old versions
             from css3-page WD. But was implemented by at least one printer
             implementation. Added section to allow such implementations to
             interpret old syntax if needed for content-compat.

   glazou: Please add Changes from Previous Version section
   fantasai: Ok, I can do that.

   <dbaron> I think that (1) if we publish the document, it should have an
            editor listed (fantasai, I think) who is an active member of
            the working group and (2) it should probably also have a public
            editor's draft if it's an active document
   glazou: Any objection to publishing?
   fantasai: I don't think it should be an active document. Think we just
             publish this update, and then ignore the fact that it exists.
   RESOLVED: Publish css-print with fantasai as editor, updated changes section.

Conditional Rules CR
--------------------

   <dbaron> http://lists.w3.org/Archives/Public/www-archive/2013Feb/0040.html
   dbaron: At F2F we had almost all issues resolved, a few left
   <dbaron> http://lists.w3.org/Archives/Public/www-style/2013Feb/0229.html
   dbaron: First one was proposal for issue 5, which is behavior of insertRule
   dbaron: I looked at what implementations do, not quite consistent.
   dbaron: Seem we like WebKit behavior best, so suggest we spec that.
           I've already implemented it in Gecko.
   dbaron: Question is basically what happens if you pass insertRule an
           empty string, or multiple rules, or valid rule with other
           garbage afterward.
   dbaron: Proposal is they all throw SyntaxError exception b/c not a
           valid single rule.

   SimonSapin: Don't we have same on stylesheet object?
   dbaron: I would expect same rules to apply there.
   dbaron: Spec was equally unclear
   ACTION: Glenn to update CSSOM to throw SyntaxError on insertRule with
                 above weirdness as argument
   <trackbot> Created ACTION-545
   <SimonSapin> should css3-syntax define what is valid?

   dbaron: Others, one issue was unclear if had addressed; had been.
   <dbaron> http://dev.w3.org/csswg/css3-conditional/doc-20121213-LCWD.txt
   dbaron: Others were editorial, plus one resolution that was missing edits.

   glazou: Colorized DoC?
   glazou: Helps for the conf call with staff
   glazou: If you don't have time, don't worry, but if do, that will help
   glazou: Any objection to move to CR?
   Florian: No, let's go!
   RESOLVED: css3-conditional to CR

   dbaron: Would like to link to test suite at time we publish CR.
   dbaron: Have a bunch of tests, but no built test suite.
   plinss: I'll take care of that.
   florian: You're referring to tests contributed by Mozilla and by me?
   dbaron: Yes, would prefer something that's more than a query in shepherd
           to refer to
   ACTION: Bert start process for CR for css3-conditional
   <trackbot> Created ACTION-546

Counter Styles LC
-----------------

   glazou: Tab, counter styles?
   TabAtkins: Edited all issues based on F2F discussion
   http://dev.w3.org/csswg/css-counter-styles/
   TabAkins: Want LC
   fantasai: we just added the new feature (for 0-filling), I think we
             should publish a WD today or so
   fantasai: and then give people a few weeks to review before LC
   <glazou> fantasai, TabAtkins, missing Changes from Last Version here too

   glazou: No objection to WD?
   fantasai: Tab, please update the changes section, and I'll deal with
             a quick publication request
   RESOLVED: New WD counter-styles, expect LC in 2 weeks or so

Multi-col Percentages
---------------------

   SimonSapin: We had proposal on mailing list to add percentages to
               column-width or column-gap
   SimonSapin: Do implementers want to do this?
   SimonSapin: Should we do this in CR?
   glazou: Have a use case for this. If you try to show an editing grid in
           bg of document, using background is very useful
   glazou: Setting columns to percentages will ensure columns map to the
           grid.
   glazou: If you try Adobe, does this.

   fantasai: Why not use column-count?
   glazou: I think it's not enough.
   TabAtkins: Seems useful
   dbaron: I think it'll confuse people into thinking it's the preferred
           way to get certain number of columns
   dbaron: It's not, because there's gaps, and things won't quite add up.
   dbaron: They will get unexpected results.
   SimonSapin: I think request was first for column-gap, then column-width
               b/c looked easy, but maybe we don't need that.
   TabAtkins: Given column-gap: <percent> use case is handled by column-count,
              ok with me
   dbaron: column-gap is fine with me, as long as we clearly say what it's
           relative to

   glazou: Ok with me too, as long as we have percent for column-gap..
   glazou: Any objection to adding that to spec?
   Florian: Which level?
   fantasai: Have to go back to LC for other edits anyway
   RESOLVED: Add percentages to column-gap, not to column-width (use
             column-count for that, it's better)

Font Load Events Host Spec
--------------------------

   Topic: Splitting font load events out of fonts spec
   jdaggett: Think one issue we can resolve quickly, font load events in
             fonts spec
   jdaggett: font-load events, important issues about [?]
   jdaggett: People leaving various comments ...
   jdaggett: Potential for churn on this one portion of the spec, and seems
             would make sense to push out to separate spec.
   jdaggett: If ppl ok with that, will take out of spec, and put together
             something else
   jdaggett: Would like resolution on pushing out font load events.
   glazou: I can live with that, no problem
   <dbaron> fine with me
   TabAtkins: I agree
   <SteveZ> OK, with removal, would like quick progress on separate document
   RESOLVED: Push font load events out to separate spec

Sylvain's Farewell
------------------

   glazou: One last thing, let's all wave goodbye to Sylvain!
   <smfr> bye sylvaing!
   <darktears> sylvaing: good bye!
   <dbaron> bye sylvaing, and thanks
   glazou: I hope you'll be around for something else, another WG in
           consortium...
   glazou: If it's the case, see you at TPAC
   <tantek> bye sylvaing! hope to see you soon.
   <hober> sylvaing: don't go! :)
   * SteveZ waving good-bye and welcome
   <bradk> bye sylvaing
   <sylvaing> bye everyone
   * sylvaing and thanks SteveZ
   <JohnJansen> boo SteveZ

Meeting closed.

[side discussion of using Presto as an implementation for CR qualification,
  partially excerpted below]
<sylvaing> tantek, i think if users can't download it it's not shipping.
            if they stop improving it but you can download it then it still
            counts though not for much longer since nothing new will happen
            there
<SteveZ> The point was that other can replicate tests using the "shipped"
          implementation.

Received on Thursday, 21 February 2013 06:03:13 UTC