W3C home > Mailing lists > Public > www-style@w3.org > April 2013

[CSSWG] Minutes Telecon 2013-04-24

From: fantasai <fantasai.lists@inkedblade.net>
Date: Wed, 24 Apr 2013 12:41:34 -0700
Message-ID: <5178356E.6070701@inkedblade.net>
To: "www-style@w3.org" <www-style@w3.org>
Summary:

   - Start adding agenda topics for Tokyo to wiki!
   - RESOLVED: publish WD of selectors 4 with note above wrt :matches()
   - Discussed publishing Template Layout spec. Week to review.
   - Discussed issue with flow-into.
   - Need info from fonts people wrt use of font metrics for alphabetic
     underline position
   - RESOLVED: Change text decoration z-ordering back to match 2.1
   - RESOLVED: Repeat position:fixed on all pages, including blank pages
               genarated by forced page breaks. Tweak the definition of
               :blank to match.
   - Case matching of fonts seems to be resolved, unless someone has concerns:
       http://lists.w3.org/Archives/Public/www-style/2013Apr/0553.html
   - Reviewed changes to CSS3 Box Alignment:
       http://dev.w3.org/csswg/css-align-3/

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

Present:
   David Baron
   Bert Bos
   Rik Cabanier
   Arron Eicholz
   Elika Etemad
   Simon Fraser
   Sylvain Galineau
   Daniel Glazman
   Koji Ishii
   Dael Jackson
   Peter Linss
   Anton Prowse
   Matt Rakow
   Florian Rivoal
   Simon Sapin
   Dirk Schulze
   Alan Stearns

<RRSAgent> logging to http://www.w3.org/2013/04/24-css-irc
Agenda: http://lists.w3.org/Archives/Public/www-style/2013Apr/0556.html
ScribeNick: dbaron

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

   glazou: I included a few extra agenda items already; saw one more from
           Tab about display:none-ness, but Agenda quite full already.
   glazou: Two first things...

   glazou: Need to start collecting agenda for Tokyo, please add items to
           wiki.
   glazou: I think we still need some admin details (e.g., hotels) from
           jdaggett.  Or maybe I missed them.
   glazou: I'll ping jdaggett to get these details.

Selectors 4 Publication
-----------------------

   glazou: We have request from fantasai to publish selectors4.
   <glazou> http://lists.w3.org/Archives/Public/www-style/2013Apr/0418.html
   fantasai: Tab and I updated a whole bunch of things in selectors4.
   fantasai: Simon Sapin sent some comments that we still need to address.
   SimonSapin: I sent a bunch of feedback, but nothing blocking WD.

   arron: We had a discussion about author requirements. That note needs
          to go in module template and all specifications, saying author
          requirements are non-normative.
   fantasai: I think they should be normative.
   arron: then we need to test them
   arron: In CSS 2.1 we had a note saying author requirements are not normative.
   sylvain: ...
   * dbaron is astounded by that text!
   arron: In css3 module template we have removed that text
   arron: Which means we assume author requirements are now normative and
          testable.
   dbaron: I don't think we've ever done it this way.
   arron: We have to test all normative requirements.
   * sgalineau thinks this is a complete non-problem
   Florian: I think it's fine for author requirements to be normative.
   dbaron: We have to test all the implementation conforming requirements.
   dbaron: We have to demonstrate that we have two conforming implementations;
           don't need to demonstrate 2 conforming authors.
   arron: How do you prove implementable by authors?
   dbaron: Not part of the process.
   dbaron: I'm strongly opposed to this text in 2.1 and didn't know it was
           there.
   arron: in section 3.1
   Florian: This undermines validators.
   <sgalineau> If the process makes it harder to specify authoring requirements
               the process needs to change
   arron: authors free to do what they want with documents...
   fantasai: So are implementations. Doesn't mean they're conforming.
   dbaron: If no author requirements are normative then we have no definition
           of a conforming document.
   arron: ... demonstrate?
   dbaron: I think we want 2 conforming implementations, ... though somebody
           should read process.
   peter: We have to meet the exit criteria we put in our own specs.
   peter: We only talk about implementations.
   arron: I don't want to push specs out that don't have this.
   arron: We need to define this more clearly, maybe in testing area.
   <oyvind> CSS 2.1 section 3.1 just says something about "recommendations",
            and that text applies to both authors and UAs

   dbaron: I wasn't sure why combinators inside :not() and :matches() was
           moved into full profile
   <fantasai> foo :matches(bar baz)
   <fantasai> <foo><bar><baz/></bar></foo>
   <fantasai> <bar><foo><baz/></foo></bar>
   fantasai: This allows branching selector path.
   fantasai: I'm fine to include it if that's what people want.
   dbaron: Already have to do that for mixes of child and descendant combinators
   <dbaron> body > div p
   dbaron: It's slow but I don't think it has totally different implementation
           characteristics.
   dbaron: Maybe put an issue in the spec saying we're not sure and people
           will comment
   fantasai: sounds ok
   RESOLVED: publish WD of selectors 4 with note above

Grid Template Layout
--------------------

   glazou: Bert asked to publish WD of grid template layout
   Bert: last WD more than a year old
   Bert: I've integrated discussion of regions, more integrated with content,
         page templates, and regions
   Bert: I think it's quite nice now.
   Bert: Not yet LC, too many details that can go either way; in some cases
         doesn't matter.
   Bert: But I think it's time for a new WD.  Lots of things to do with page
         layout at the moment.
   Bert: Would be nice to have something new in that area.
   glazou: I have a problem with this document.  Because it's a challenging
           spec for the grid layout that is implemented by browser vendors,
           and I don't see browsers expressing interest in this one.
   glazou: I'd like to hear from browser vendors on this call:  are you
           interested in this spec, the different syntax it proposes for grid?
   glazou: because apparently nobody's implementing it
   Bert: But the existence of prototypes outside of browsers in indication
         people want it.
   glazou: One from CÚsar, quite old.
   Bert: CÚsar made two, and there are two others.
   glazou: Polyfill or native implementation?
   Bert: 3 JS, one native
   glazou: I'd like at least one week to review the document.
   glazou: It seems to me another content sink for stuff that should probably
           go elsewhere.  What do other people think?
   dbaron: I haven't had a chance to follow it.
   florian: same here
   <florian> s/?/florian/
   fantasai: Haven't looked at most recent WD, but found that while I don't
             think Bert's module would go to LC as such, has been able to
             pull together ideas from different specs and see how they
             integrate.  Reasonably useful thing for someone to do.
   glazou: Seems we need some time to review the document.
   glazou: Bert, is 1 week for review ok?
   Bert: 1 or 2 weeks is fine.  As I said, a year old.
   glazou: ok, will revisit in 1 or 2 weeks
   ACTION all: review grid template layout

Color
-----

   Topic: color-correction into color-interpolation
   <stearns> http://lists.w3.org/Archives/Public/www-style/2013Apr/0096.html
   dbaron: Didn't we do this last week?
   dirk: I didn't put on agenda; discussion continuing on mailing list.

Multi-Column Layout
-------------------

   Topic: column-fill and column-overflow
   <glazou> http://lists.w3.org/Archives/Public/www-style/2013Mar/0042.html
   dbaron: Don't think it needs telecon time now.

Regions
-------

   Topic: flow-into
   <glazou> http://lists.w3.org/Archives/Public/www-style/2013Apr/0181.html
   <stearns> http://lists.w3.org/Archives/Public/www-style/2013Apr/0471.html
   Stearns: Has been on agenda for a few weeks, I just went ahead and did it.
            Second message describes change to region spec.
   Stearns: Don't know if we need time on the call unless people want to
            comment.
   Bert: Initial reaction: dangerous, sounds like XSLT territory redoing
         the tree.
   Bert: Want to see examples of what it's useful for.
   Bert: And want to try doing those things different ways without
         destroying the tree.
   glazou: Needed at least for selectors not being able to target text nodes.
   glazou: So using combinator with a star not enough
   Bert: Have example in css3 layout spec
   Bert: Have example ...
   Bert: worried it's not quite what we want
   Bert: ... part of the content, not all of the content ....
   Stearns: Bert, would definitely like to see use case
   Stearns: Initially came about because Rossen was asking for it, maybe
            Rossen can post use cases
   Stearns: In general, going back into history of regions, this is
            something that was discussed very early on
   Stearns: alexmog had action to figure out way of doing this correctly
   Stearns: earlier had decided to push to level 4, decided to pull back
            to current level
   Bert: what happens to inheritance if element removed
   Stearns: as with all named flow mechanisms not modifying dom tree
   Stearns: All we're doing is taking the boxes in element's content area
            and moving them to namedd flow and displaying in region chain
   Bert: maybe we should not discuss here, might take too much time.
         But all kinds of difficult cases.  Though maybe I don't understand
         what you want... not sure if there's really a problem.
   Stearns: would like to see difficult cases on list
   glazou: so, move to mailing list

CSS3 Text
---------

   Topic: Recommended justification for symbols
   <glazou> http://lists.w3.org/Archives/Public/www-style/2013Apr/0331.html
   Koji: you say I should make a recommendation?
   fantasai: If nobody else has an opinion I think Koji and I can handle
             it offline

CSS3 Text Decoration
--------------------

   Topic: Text decoration issues
   <glazou> http://dev.w3.org/csswg/css-text-decor-3/issues-lc-2013#issue-6
   fantasai: first is from dbaron, about dropping
             text-underline-position: alphabetic
   <glazou> http://lists.w3.org/Archives/Public/www-style/2013Mar/0529.html
   fantasai: your issue is that alphabetic is requesting to use font data
             except when it's incorrect
   dbaron: My big problem with this is that I don't know how it would really
           be implemented
   fantasai: I'd like to hear from jdaggett.
   fantasai: He knows a lot about font metrics and what's an appropriate
             way to use them
   glazou: fantasai, can you take action to ping jdaggett
   fantasai: did ping, haven't heard back
   glazou: maybe handle off call?

   fantasai: does anyone else have feedback on issue? Esp. Adobe?

   glazou: ok, issue 8
   <glazou> http://lists.w3.org/Archives/Public/www-style/2013Mar/0526.html
   fantasai: dbaron pointed out z-order is opposite of what's in 2.1
   fantasai: changed underlines/overlines to draw over text
   fantasai: I think this is an unintentional change
   fantasai: Anybody have any idea why it might have been intentional?
   <silence>
   fantasai: OK, I'll chnge that back to match 2.1
   RESOLVED: text decoration z-ordering back to match 2.1

   glazou: anything more to talk about?
   fantasai: need to just sit down and clarify stuff

    * sgalineau for some reason thought overline/underline order was related
                to text-shadow
   <BradK> sgalineau: Text shadow above the underline might look weird.
   <sgalineau> bradk: yeah. I can't recall the reasoning but I thought
               we discussed the paint order of these when we debated
               text-shadow. could be wrong...
CSS3 Paged Media
----------------

   Topic: fixed positioning on content empty pages
   <glazou> http://lists.w3.org/Archives/Public/www-style/2013Apr/0546.html
   SimonSapin: So there's 1 sentence in CSS21 that says fixed pos is
               repeated on every page in paged media.
   SimonSapin: Do we want to repeat in empty pages that are generated by
               page-break: left/right?
   SimonSapin: And do such pages match the :blank selector?
   fantasai: Only pages that should match :blank are ones that are generated
             by forced breaks
   fantasai: If you have a very tall empty block and it prints across 2
             pages...
   fantasai: Don't think it makes sense for that to match blank.
   fantasai: might have a background
   SimonSapin: so we should tweak dfn of :blank or content-empty?

   fantasai: What would be more useful?
   fantasai: I think only forced break pages matching :blank would be more
             useful.
   SimonSapin: 2 options:
   SimonSapin: (1) repeat fixed position boxes on every page including pages
                   generated by forced breaks
   SimonSapin: (2) not display fixed boxes on pages that are supposed to
                   be blank
   SimonSapin: And in any case tweak the definition of :blank and content-empty
               so it's what we want.
   fantasai: I have no opinion on this.
   fantasai: Maybe ask murakami-san
   <dbaron> I think break: right/left is kinda like forced breaks btw.
   Bert: I don't know what :blank should match, but I do think fixed elements
         should repeat on all pages.
   Bert: I don't see why there should be page number there but no fixed
         content. (???)
   Bert, is that what you said?
   SimonSapin: That's what I prefer too.
   SimonSapin: In that case we need to make sure :blank still applies on
               such pages.
   SimonSapin: I can take an action to edit that in the spec.
   ACTION Sapin: Clarify definition of :blank selector in relation to fixed
                 positioning
   <trackbot> Created ACTION-557
   RESOLVED: Repeat position:fixed on all pages, including blank pages
             genarated by forced page breaks. Tweak the definition of
             :blank to match.

Font Name Case-matching
-----------------------

   glazou: Not sure if there's something to decide about unicode caseless
           matching.
   <glazou> http://lists.w3.org/Archives/Public/www-style/2013Apr/0553.html
   glazou: Message from Addison Phillips about unicode caseless matching.
   glazou: Want to make sure all members have seen it.
   glazou: we finally got an answer to our questions
   glazou: jdaggett not on call, not sure discuss now, but will probably
           have to discuss to make sure we're in sync
   glazou: But if someone has a comment...
   Bert: If it's indeed safe to ignore normalization.  Question of seeing
         what different OSes do.  Sounds dangerous to me.

CSS3 Box Alignment
------------------

   <glazou> https://lists.w3.org/Archives/Member/w3c-css-wg/2013AprJun/0069.html
   fantasai: We ran out of other topics?  Is Tab on the call?
   <fantasai> http://dev.w3.org/csswg/css-align-3/
   fantasai: So we changed a bunch of things in this spec.

   <fantasai> http://dev.w3.org/csswg/css-align-3/#alignment-values
   fantasai: Pulled out values and defined them in a common seciton.
   fantasai: Two new values, self-start and self-end
   fantasai: Came across the issue of whether start and end are determined
             w.r.t. container or the box's own writing mode
   fantasai: overflow handling in CSS 2.1 handled by looking at containing
             block's direction, so decided containing block direction should
             be default interpretation of start and end
   fantasai: But sometimes want to align by box's own writing mode, so added
             self-start and self-end.
   fantasai: An example would be grid, column of stuff with most l-to-r,
             some r-to-l, want to align all to start-edge (all left).
   fantasai: Some specific cases might want based on writing mode,
             self-start and self-end address that.

   fantasai: Second, we defined baseline alignment.
   fantasai: Ton of text added to do this.  We think it's right, but probably
             will be a while before it's all right.

   fantasai: In old draft there was a 'true' keyword that was optional.
   fantasai: In new draft we have 'true' and 'safe' keywords.
   fantasai: And if neither is used, default depends on layout mode.
   fantasai: Document-centric layout modes, defaults to 'safe'.
             (Matches CSS2.1.)
   fantasai: For design-centric modes, defaults to 'true'. (Matches Flexbox.)

   fantasai: Next, can combine content distribution keyword e.g. space-around
             with position keyword e.g. start/end
   fantasai: Says what to do if you only have 1 thing.
   fantasai: Added another content distribution keyword: space-evenly
             (like border-image) added to space-between and space-around
             (in flexbox).
   fantasai: That came up in discussion we had about flexbox.
   fantasai: We'd only added 2 for flexbox, but third might be useful for
             grid.
   fantasai: E.g., if stuff on sides is not grid.

   fantasai: Worked a bit on defining details of how things align.
   <fantasai> http://dev.w3.org/csswg/css-align-3/#content-distribution
   fantasai: Details of how justify-content...
   fantasai: section talks about block containers, (reads)
   fantasai: exlpains how vertical-align continues to work on table cells
             by default

   fantasai: we also talked about self alignment
   fantasai, tricky bit was abspos and replaced elements
   <fantasai> http://dev.w3.org/csswg/css-align-3/#justify-self-property
   fantasai: we have stretch keyword... block stretches by default if
             non-replaced
   fantasai: but if replaced or has width, replaced aligns to start
   fantasai: so stretch and start are equivalent on block level boxes
   fantasai: for abs pos we envision alignment ... take containing block,
             offsets reduce cb, and within adjusted box normally behavior
             is stretch if you have auto w/h
   fantasai: but if you take some other keywords but want to make it easy
             to center abs pos as well
   <dbaron> This doesn't sound 2.1-compatible to me.
   fantasai: If alignment property is not auto or stretch then you shrink-wrap
             and align with it instead of trying to fill whole box.
   fantasai: Replaced elements don't do that.
   fantasai: So for replaced element stretch keyword is equivalent to start.
   fantasai: Trying to figure out if way to allow replaced elements to stretch.

   fantasai: We added to justify-items property
   <fantasai> http://dev.w3.org/csswg/css-align-3/#justify-items
   fantasai: we wanted to handle center elements and the html align property
             somehow
   fantasai: and came up with using justify-items property and adding a
             special keyword, and if that keyword is present
   fantasai: auto value of justify-items checks parent and if parent includes
             legacy keyword copies value to self
   fantasai: Was best we could come up with in terms of behavior that would
             handle that.
   <dbaron> What about my old proposal?
   fantasai: If anybody has better idea about how to handle html stuff,
             open to better ideas?
   fantasai: And do we want it to be called legacy?  legacy of 1990, legacy
             of parent?
   fantasai: Currently only allowed combination with left/right/center.
   fantasai: Potentially useful to authors to use this thing, in which case
             do we want to allow combination with other kewyords?

   fantasai: That's the overview of what we've done.
   fantasai: It's going to need another few rounds of review edit
   fantasai: Want to publish WD at some point but not high priority if people
             want time to look at things, week, 2 months, whatever?
   <dbaron> I want some time to look at things
   <dbaron> Is there a written summary of what you just described somewhere?
   fantasai: No written summary, but could create one today.

   Bert: Had three remarks.
   Bert: So far we've used center for horiz and middle for vertical
   Bert: This only uses center, maybe should use middle
   fantasai: Actually we don't have that distinction, if you look at
             background-position, it uses center for both
   fantasai: and vertical-align: middle dosen't do true centering... does
             weird text-middle thing
   <dbaron> does for tables!
   fantasai: We wanted to avoid confusing things with the text-middle behavior
             and be consistent with bg-pos.
   Bert: For tables middle really is center

   Bert: I don't like introducing extra property for true centering.
         We broke margin:auto
   Bert: Don't like introducing property that changes meaning of another.
   Bert: want margin: fill that means what margin auto used to mean
   Bert: I know we did box-sizing in the indirect way
   dbaron: This seems a lot less indirect to me in that it is doing block
           centering, that kicks in when there are no auto margins
   dbaron: Authors don't like working with margin: auto
   dbaron: We have a model where you can do both. You can use margin: auto.
           but if it's not there, alignment properties can be used
   dbaron: Isn't the true value on the same property as these alignment
           properties?
   fantasai: yes, it's an optional keyword
   Bert: oh, a shorthand?
   fantasai: Not a shorthand, but takes multiple keywords

   Bert: third remark:  have a stretch/distribute value to justify blocks
         inside container
   Bert: space, I think
   Bert: But I'm missing... use case wanted all blocks to be just one after
         the other except the last block that should go all the way to the
         other end
   Bert: cannot do that with spacing between
   Bert: I wanted something like an auto margin
   fantasai: I think case would have to be handled with margins... is in
             flexbox with auto margins
   Bert: I was hoping to see that outside floxbox
   Bert: I want it outside flexbox; flexbox has no collapsing margins.
         Author name at bottom.
   fantasai: I think reasonable request, has to be handled in spec defining
             margins on block
   fantasai: might want to, for example, push last two elements, not just
             last one, to the bottom.
   fantasai: We'd have to add a new kewyord, but I think that's out of scope
             for what we're doing here.

   fantasai: When to publish WD?  When to discuss?
   <dbaron> I'd like some time to review
   Bert: I'm fine with publishing now.
   dbaron: I'd like at least a week to review
   glazou: so revisit next week

Meeting closed.
Received on Wednesday, 24 April 2013 19:42:04 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 24 April 2013 19:42:04 UTC