[CSSWG] Minutes Telecon 2013-12-11

September F2F
-------------

  - The September F2F will be 8-10 September unless there are objections
         raised soon.

Compositing and Blending
------------------------

  - Cabanier will make edits regarding functionality for the group to
         review before next week's call.

CSS Masking to CR
-----------------

  - Krit requested more time to go over edits he recently received.

WebVTT Review
-------------

  - ChrisL had some comments and suggestions that he will send to the
         VTT group.
  - Other members of the WG will continue to review and send comments
         later if they have them.
  - Everyone appreciated being brought into the process early on.

Will-Animate Proposal
---------------------

  - Several different concerns about the proposal were raised, including
         giving too much access and how exactly it works with stacking
         and animations.
  - A summary of the current status of the proposal will be posted to
         the mailing list in order to further conversation.

Interpolate() Proposal
----------------------

  - Discussion was deferred.

:Sorted Pseudoclass
-------------------

 - RESOLVED: Add :sorted pseudoclass to selectors 4

=====FULL MINUTES BELOW======

Present:
  Glenn Adams
  Rossen Atanassov
  David Baron
  Bert Bos
  Rik Cabanier
  Tantek �elik
  Dave Cramer
  Elika Etemad
  Simon Fraser
  Sylvain Galineau
  Daniel Glazman
  Dael Jackson
  Philippe Le H�garet
  Chris Lilley
  Peter Linss
  Anton Prowse
  Florian Rivoal
  Simon Sapin
  Dirk Schultz
  Alan Stearns
  Steve Zilles

Regrets:
  Tab Atkins
  Mihain Balan
  Brad Kemper
  Kang-Hao (Kenny) Lu
  Simon Pieters
  Leif Arne Storset
  Lea Verou

scribenick: dael

  plinss Let's get started
  plinss: Any additional items?
  sylvaing: Please put your name for registration for the January F2F.
  sylvaing: It's especially important because we need to issue badges so
            you can get in.
  sylvaing: The earlier a complete count is available the better.
  * plh plans to come
  * ChrisL will be there and has planes booked
  plinss: This is a good time to make travel arrangements.

September F2F
-------------

  SteveZ: I believe the AB meeting is the 16-17th Sept
  SteveZ: I think Bert was offering to host.
  SteveZ: Optimally if it was near those dates I would only need one
          trip to Europe.

  krit: I would prefer after.
  glazou: Me too.

  plinss: Any other preferences?
  sylvaing: With TPAC afterwards, it was inconvenient that summer,
            autumn, and winter meetings were close together.
  <Bert> (TPAC is Oct 27-31)
  plinss: That is true.
  plinss: The week of the 22nd would be 4 weeks before
  krit: If we did the week before that would be okay.
  SteveZ: That's fine with me.

  plinss: Bert, are any these dates a problem?
  Krit: TPAC is on Halloween so it must be in October.

  plinss: So do the week of 8-10 September?
  plinss: That work for everyone?

  ChrisL: That's fine for me.
  florian: If we move further from TPAC it's better.

  plinss: Okay, unless we hear other complaints soon, let's call it
          September 8-10.

Compositing and Blending
------------------------

  cabanier: I got comments from James Robinson,
  cabanier: He's asking for something that doesn't require spec changes,
  cabanier: Should I wait another week to ask for CR?
  cabanier: Or does that not stop it?

  ChrisL: It's obviously a judgement call.
  ChrisL: If it's ongoing and going to produce significant changes you
          should wait,
  ChrisL: If not you can push it to later.

  cabanier: He's asking for clarification on something in the spec.
  ChrisL: It's just a clarification you're fine.
  cabanier: Should I put it in disposition of comments?
  ChrisL: Yes.

  ChrisL: Is there something we can look at?
  cabanier: Yes, I'll paste in IRC.
  <cabanier> doc: http://dev.w3.org/fxtf/compositing-1/issues-lc-2013.html

  ChrisL: One comment, introduce some classes there with colors,
  ChrisL: So someone can see green with agreement etc. That makes it
          easier.
  ChrisL: You should distinguish between where everyone agreed and
          everyone talked about it and the person is happy even if they
          didn't get want they wanted.
  cabanier: Okay.

  <fantasai> http://dev.w3.org/csswg/css-flexbox/issues-cr-2012
  <fantasai> http://dev.w3.org/csswg/bin/issuegen.pl
  fantasai: First link above is an example of a color coded one.
  fantasai: And the 2nd is a template to generate from text.
  cabanier: That's handy.
  fantasai: If you do this it spits out instructions and it'll create
            the HTML.
  fantasai: There's others you can use, this is the one I wrote.
  cabanier: I was surprised when Cameron told me I had to register.
  cabanier: I'll update.

  <dbaron> I'm a little concerned about the number of parts of the spec
           marked non-normative.

  cabanier: Should I wait for CR is it it fine?
  plinss: As long as it's updated for the spec. call I think it's okay.
  cabanier: The comments from James will be in there.

  smfr: I find the spec confusing because it has a lot but doesn't
         provide new properties.
  smfr: For example it has a section on knockout groups, but it doesn't
        have examples.
  smfr: It's very confusing because it has a long section on theory and
        I don't know why it's relevant.
  cabanier: It has a lot of normative text.
  <dbaron> There are a bunch of references to a 'knockout' keyword
           that's no longer in the draft.

  smfr: I think technical information is fine, they just want to know if
        they can use it now, use it later, what's being developed.
  smfr: It's explaining what it's for.
  smfr: Does this need knockout groups?
  smfr: There isn't an SVG for this, right?
  cabanier: I don't think anyone implemented that.
  smfr: This looks theoretical.
  smfr: I think it should stay.

  dbaron: I don't think so. I think there's a bunch of references to
          something that as there before.

  smfr: Is that for level 2?
  smfr: Knockout we've tried for some time, but if it's to level 2
        that's okay.

  sgalineau: I think there's two options. 1: We remove knockout. 2: We
             make clear which isn't covered.
  cabanier: I think we can remove things not covered. I think it can be
            done in CR.
  cabanier: That second section is informative about how it could work.

  smfr: We're removing things not covered so it's confusing,
  smfr: And having a long normative section is dangerous.

  dbaron: I think it's not clear what's non-normative.
  dbaron: Does this section mean what is non normative. Like is 5
          non-normative, but 5.1 isn't?
  cabanier: I should be in the header.
  sgalineau: It should be an appendix.

  dbaron: To be clear sections 5, 6, 7, 8, 9, and 10 are marked as
          normative.
  dbaron: Does that mean sections 5, 6, 7, 8, 9, 10 don't define
          anything but property syntax?
  cabanier: We discussed this and at the time only properties were
            supposed to be marked as non-normative.
  dbaron: Who told you that?
  cabanier: I don't remember. I can look.

  plinss: Unless there's a good reason, that's the sort of thing that
          should be normative.
  plinss: You shouldn't have someone implement a blend mode and get
          different results.

  smfr: It sounds like that should change in the spec. It's important.
  * sgalineau thinks the backgrounder should be an appendix.

  plinss: Are we in agreement those sections should be normative?
  ???: We should make it normative in a later version.
  plinss: I think we agreed earlier that items not referenced should
          move to the next level.
  plinss: We should shift text from not normative to normative before
          CR.
  plinss: Do we want to see these changes and then revisit going to CR
          or resolve now and see it later
  smfr: I think it was clear, but some people found it confusing so
        should we wait to see if this makes it less confusing?
  ???: This is something we're going to remove is CR, so why not do it
       now?
  cabanier: We can do it next week.
  ???: Otherwise it looks like we're removing for no reason.
  cabanier: I wanted to be less confusing in LC. I think I'll make the
            changes and discuss next week.
  plinss: Agreed. The only problem is editing functionality.
  ???: We're not changing any behavior.
  plinss: So you have your orders for changes and we'll revisit next
          week.

CSS Masking to CR
-----------------

  krit: I got comments a few hours ago.
  krit: I got fantasai's comments.
  krit: Her comments will delay CR, so I'd like to discuss on the
        mailing list,
  krit: Since her requests are changing behavior and names.
  plinss: So you want that before CR?
  krit: Yes.
  plinss: Let us know when you want to revisit.

WebVTT Review
-------------

  plinss: Everyone should have reviewed for feedback.
  Bert: I promised to review
  ChrisL: I have some comments, Bert can go first.

  plinss: I heard one person ask for time and one person had feedback.
  ChrisL: I had a little.
  ChrisL: WebVTT specifies certain properties must be set to particular
          values,
  ChrisL: And lists the properties that must apply (the others must be
          ignored).
  ChrisL: Is it correct that there is no DOM interface to get at the
          styling information, so the only way to see what properties
          are applied is if
  ChrisL: They have a particular visual effect?

  plinss: Do you have a suggestion for a better way to phrase that?
  ChrisL: I think I'm looking for a change in words.
  ChrisL: Saying you can't do these things is odd.
  ChrisL: It's better to say should not instead of must not.
  ChrisL: If you say must not you have to test and there's no way to do
          that.

  plinss: Okay, any other feedback?
  plinss: I hear folks need more time to review.
  israelh: I could do with more time for reading it.

  ChrisL: I have a follow-up.
  ChrisL: If we want tests for this, currently we can do SVG and HTML
          tests, can we integrate WebVTT?
  plinss: I don't understand how it would be significantly different.
  ChrisL: I was asking you.
  plinss: I think it's fine
  <ChrisL> ok cool

  israelh: Would you be able to share the location again in IRC?
  <ChrisL> https://dvcs.w3.org/hg/text-tracks/raw-file/default/webvtt/Overview.html

  glenn: Just as a note, VTT is a community group and hasn't entered
         into W3C rec track.
  glenn: It also doesn't have a status which is required before rec
         track.
  glenn: Current plan to to take it through the timed text WG.
  glenn: I wanted to mention that comments might be a bit premature.
  ChrisL: I understood, but I think it's good they asked for feedback
          earlier;
  ChrisL: I think they should be commended for asking early.
  plinss: Agreed.

  plinss: I'm not hearing anything else, but that concerns me.
  plinss: Do people need more time or are they okay?
  SteveZ: I think where they are would allow us to add more comments as
          we find them.
  plinss: I just want to make sure it happens.

  plinss: Only question is, all we have is Chris's comments. Do you want
          to send that yourself?
  ChrisL: Yes. I think it would be good to have something from the
          chairs saying there we discussed and there may be more
          comments later. Tell them they're good for asking.
  plinss: Yes.

Will-Animate Proposal
---------------------

  plinss: Anyone want to speak about it?

  smfr: I can summarize.
  smfr: As far as I can understand, it will allow authors to say later
        they'll change something with an animation.
  smfr: This is a trigger to allow the user agent to prepare for an
        animation to start.
  smfr: The engine may create something ahead to allow it to work more
        smoothly.
  smfr: This is exposing implementation detail and may change in future
        to make this unnecessary or confusing.
  smfr: I'm not a big fan and I think this could be mis-used to force UI
        to allocate too much memory.

  dbaron: There were some design details to avoid exposing too much
          detail.
  dbaron: That said, one of the problems is there are a bunch of
          properties where everything by default causes stacking.
  dbaron: There's a desire for will-animate to cause that even without a
          new value.

  SimonSapin: I want to clarify that this proposal works with stacking
              behavior.
  smfr: Right, and authors can create stacking now.
  smfr: I don't see the need for will-animate to do that.
  smfr: Is the desire so that later on you create less work for the UI
        to do?
  dbaron: Yes.
  SimonSapin: This proposal creates stacking contexts, so it's not just
              about performance but also affects behavior.

  smfr: In webkit creating a stacking context isn't a lot of work.
  dbaron: The work is around creating layerizing. When it's not stacking
          it has to layerize differently.
  smfr: I think that's different between UIs.
  dbaron: I don't think it is. I think it has to do things in a way that
          makes it inherently more expensive.

  plinss: I also have a lot of concerns for similar reasons.
  plinss: It's the wrong place for optimization.
  plinss: What concerns me is adding the stacking. It may be useful
          because you don't want layering to change with animation, but
          I'm not seeing that happening.

  dbaron: I think it's worth talking about why we want this.
  dbaron: There are cases an author knows they want to animate,
  dbaron: For example touch interface.
  dbaron: When the user touches they know it'll move,
  dbaron: It's important that the response is instant.
  dbaron: When we're talking about trying to do touch UI on mobile
          hardware,
  dbaron: The cost of painting into a layer when it wasn't before is
          expensive,
  dbaron: And the cost of optimistically using lots of layers is
          expensive,
  dbaron: This is a hint.
  dbaron: Other then the change in stacking, it doesn't have normative
          requirements.
  dbaron: We haven't been able to get UI with touch to be responsive
          without this.
  dbaron: We need some way to address it and this is the least-bad so
          far.

  plinss: My concern is that this is changing behavior, even though you
          say it's a hint.
  dbaron: That's the problem we wanted a hint and this is what we had to
          do.
  dbaron: Authors are doing worse things right now.
  dbaron: They set translate in advance.
  dbaron: I think having something more explicite is better then
          widespread use of hints.
  dbaron: Such as things are faster in webkit if you stick translate Z,
  dbaron: That's the world we're in.

  florian: Should we create something meant to be a hint, or should we
           create a property that's also useful?
  ChrisL: That seems better to me.
  dbaron: That does create a hint, but it creates separate ways for
          spec. properties

  plinss: I do agree that is the desire is to create stacking context a
          property is better.
  dbaron: The desire isn't to create stacking context, the desire is
          animate to preferences better.

  smfr: I'm agreeing with dbaron.
  smfr: I also think that just creating stacking isn't enough for us to
        effectively run an animation without a hiccup.
  dbaron: It's not enough, but it's a good side effect.
  smfr: I think that's fine.

  plinss: Does it always create stacking, or only with some properties?
  dbaron: I don't know.
  plinss: I'd like to see stacking be a modification of another property
          and see if it needs to be explicit.
  plinss: I'd like to get rid of the hacks.

  dbaron: I think requirements; it's two properties making people set
          two properties.
  Rossen: On the other hand, with everything that's stacking today you'd
          be giving the illusion that you'll be able to put everything
          in context with a separate property.

  Florian: So everything that creates a stacking context with auto means
           that auto just is computed as force.
  Florian: You could do that without an opportunity for turning it off.

  Rossen: Like Simon pointed out, we're trying to find something
          animatable in it's proper context.
  Rossen: I particularly favor a separate property instead of stacking.

  plinss: My other question is, isn't there some way to look ahead
          through existing style and guess.
  plinss: If the issue is we're looking for something in Javascript it's
          an API not a property.

  Rossen: The animation is run through CSS not API.
  Florian: That's odd. If it's in CSS we should figure it out.
  Florian: If it's because there's Javascript in the middle we should
           address that.
  smfr: If you're using CSS we should fix it.

  dbaron: The other problem is that if authors want to rely on this they
          want to know heuristics.
  dbaron: I think it's good to give authors reliability in what
          performance they can expect,
  dbaron: And not say it's undefined.
  florian: It still feels like this is close to a property,
  florian: That says do what I say, but faster.
  * sgalineau agrees that if the impact of this property is unclear or
              varies across browsers its usefulness is questionable.
  smfr: It's saying prepare yourself for a future CSS property change.

  Rossen_: Does anyone know if web animations spec is trying to address
           this?
  smfr: I don't think so.

  smfr: I'm happy for this to continue, if someone on mailing list
        summarizes current proposal state.

  plinss: Can someone write up summary?
  plinss: And post it to the mailing list?
  dbaron: I can poke someone and see if they can.
  plinss: Then discussion will continue on the mailing list.

interpolate() proposal
----------------------

  plinss: Leaverou had a proposal.
  <glazou> She is not here.
  plinss: None of the parties are here. Should we defer?

:Sorted Pseudoclass
-------------------

  plinss: This was from Tab and I think we can discuss here.
  plinss: The proposal is to add a psudeoclass adding HTML sorting
          model.
  plinss: Anyone have thoughts?

  <tantek> would this also apply to <ol reversed> ?

  glazou: I made a comment on mailing list because I think the current
          proposal is not enough.
  glazou: It doesn't deal with columns that you don't sort,
  glazou: If you have a list of items with an index and you want that
          index to remain ordered, it won't deal with that.
  glazou: I think it's a good start to match HTML5 and we need to extend
          to be complete.
  glazou: It's something web authors are doing more and more. We need a
          way to present to the viewer.
  <sgalineau> Doesn't understand; if you leave your table as-is then it
              can't be re-sorted, right?

  plinss: My question is that we need to extend sorting model of HTML,
          not CSS.
  plinss: Are you saying CSS should override the display of the table?
  glazou: The columns sorted with select a column if it's sorted in
          another place.
  glazou: We can't select if it's not sorted.
  plinss: Wouldn't that be column-not sorted?
  <tantek> :sorted does not alter the sorting of the table/column/list
  <plinss> :not(:sorted)
  glazou: That's not possible.
  glazou: It's a bit verbose.
  plinss: The proposal was just about if it's sorted or not.

  glazou: I had another comment about the argument that is for
          determining only in index.
  glazou: We may want a way to extend to range.
  glazou: If you sort on multiple columns you may want to sort all
          without selecting the individual.

  glazou: Other then that I think we should continue. It's needed work.

  plinss: Any objections to adding this to selectors 4?

  bert: You would want to know if the table is sorted.
  bert: You can do that with subject selector, but may be easier to have
        pseudo on the whole table.

  tantek: The only comment is I had was should that apply to ordered
          lists too, other then normally ordered?
  plinss: That's an interesting point. I don't see why not.

  RESOLVED: Add :sorted to selectors 4

  Action: glazou add :sorted to selectors 4
  * trackbot is creating a new ACTION.
  * RRSAgent records action 4
  <trackbot> Created ACTION-602 - Add :sorted to selectors 4 [on Daniel
             Glazman - due 2013-12-18].

[End of Meeting]

Received on Thursday, 12 December 2013 02:46:18 UTC