[CSSWG] Minutes F2F 2009-06-04 Part I: GCPM, Flexbox

   David Baron
   Bert Bos
   John Daggett
   Arron Eicholz
   Elika Etemad
   Daniel Glazman
   Sylvain Galineau
   Molly Holzschlag
   Håkon Wium Lie
   Chris Lilley
   Alex Mogilevsky
   David Singer (partial, via telephone)
   Anne van Kesteren
   Steve Zilles

<RRSAgent> logging to http://www.w3.org/2009/06/04-CSS-irc

Scribe: Sylvain

   howcome: at last f2f, changes to the draft included: image res property,
            list of features for WD
   howcome: i proposed to drop anything that has to do with moving elements
   howcome: floating into footnotes remain
   howcome: moving elements may conflict with generated content
   howcome: what's left now is mostly implemented by two UAs
   howcome: Prince and Antenna House
   (howcome showing updated GCPM spec)
   howcome: dropping moving elements e.g. target-text
   howcome: one issue is what do we do with advanced multicolumn features
            since no one has implemented it

   discussion of CMYK and whether it makes sense for publishing
   howcome: this is customer requirement to produce PDFs for instance
   chrisl: this should really be called device-cmyk
   chrisl: there is a standard to specify color profiles
   ACTION: ChrisL review GCPM CMYK (section 16) and suggest changes/improvements

   howcome: note that character substitution (text-replace) "applies only to batch processors"
   <dbaron> http://dev.w3.org/csswg/css3-gcpm/#text-replace
   szilles, others: what is a batch processor ?
   glazman, chrisl: non-interactive agent
   howcome: note that "This property is evaluated after the content property
            and before text-transform....it is applied after the white-space
   glazman: this is useful for localization. but is it style ?
   glazman: why not keep the feature, remove the batch processor qualifier
            and put it at risk ?
   glazou: alternatively, we need a profile or a statement that interactive
           rendering engines are not required to support the feature
   fantasai: I don't think it belongs to CSS. it changes content.
   glazou: if not then the content property is not a style property either
   chrisl, howcome: this property does not alter the DOM
   <dbaron> "This property is applied after the ‘white-space’ property."
            needs to be defined much more carefully, since 'white-space'
            applies at multiple stages of the processing model
   ACTION: howcome to rewrite scope of character substitution; from batch
           processors to not requiring support from interactive rendering engines
   (discussion follows of text processing issues deemed beyond the grasp
   of 'normal people'; minute-taker assumes he's one of them)
   discussion about a more complete cascading transformation language that
     would be more useful to printers and drop the feature from the WD
   glazman: this is not a stylistic feature and therefore belongs outside
            the specification
   glazman: we can keep as is and scope it to make it optional for
           interactive rendering engines; or we drop the section from GCPM
   howcome: i'm not comfortable with dropping something that is both
            pragmatic and useful to printers today
   dbaron: as an inherited property, i'm not sure it'd make sense to use it
           more than once per document
   glazman: which brings us some other issues about the feature e.g. whether
            it's inherited or not
   fantasai: the main issue is that the feature as specified is is
             incomplete/insufficient and will be hard to complete
             within the context of CSS
   glazou: third option for the property is to scope it to @media print
   anne: does this mean browsers have to implement it for print ?
   <anne> [apparently yes]
   vote: dbaron #2, chrisl #2, arronei #2, elika #2, molly abstains,
         hakon not 2, alex #2, anne abstains, bert #2, dglazman #2,
         szilles #2, sylvaing #3
   options were
   #1 scope feature to exclude rendering engines
   #2 drop feature
   #3 like #1 + restrict to @media print
   alexmog: my problem with this doesn't have anything to do with what
            it does or why but the state of the feature wrt the rest of
            the spec. I can't implement it as is and I'd rather implement
            a spec completely
   RESOLVED: character substitution is dropped out of the GCPM working draft

   WG proceeds with quick review of GCPM sections to decide its publication
     as a WD
   <dbaron> Hmm, now that we have leaders and leading I wonder if we could
            get some other variants of either of the base words so we can
            confuse ourselves more. :-)
   fantasai: two issues about footnotes. 1) steve's comment about multicolumn
             footnotes and 2) what happens to footnotes in scrolling media
   fantasai: specifically, where is the footnote box in the document in that case?
   <glazou> (ongoing discussion about css3-gcpm)
   renaming border-parts to border-clip
   ACTION: howcome to compare and reconcile GCPM border-clip syntax with

   <dbaron> 00B9;SUPERSCRIPT ONE;No;0;EN;<super> 0031;;1;1;N;SUPERSCRIPT DIGIT ONE;;;;
   <dbaron> 00B2;SUPERSCRIPT TWO;No;0;EN;<super> 0032;;2;2;N;SUPERSCRIPT DIGIT TWO;;;;
   <dbaron> 00B3;SUPERSCRIPT THREE;No;0;EN;<super> 0033;;3;3;N;SUPERSCRIPT DIGIT THREE;;;;
   <dbaron> 2074;SUPERSCRIPT FOUR;No;0;EN;<super> 0034;;4;4;N;SUPERSCRIPT DIGIT FOUR;;;;
   <dbaron> 2075;SUPERSCRIPT FIVE;No;0;EN;<super> 0035;;5;5;N;SUPERSCRIPT DIGIT FIVE;;;;
   <dbaron> 2076;SUPERSCRIPT SIX;No;0;EN;<super> 0036;;6;6;N;SUPERSCRIPT DIGIT SIX;;;;
   <dbaron> 2077;SUPERSCRIPT SEVEN;No;0;EN;<super> 0037;;7;7;N;SUPERSCRIPT DIGIT SEVEN;;;;
   <dbaron> 2078;SUPERSCRIPT EIGHT;No;0;EN;<super> 0038;;8;8;N;SUPERSCRIPT DIGIT EIGHT;;;;
   <dbaron> 2079;SUPERSCRIPT NINE;No;0;EN;<super> 0039;;9;9;N;SUPERSCRIPT DIGIT NINE;;;;
   <dbaron> 2070;SUPERSCRIPT ZERO;No;0;EN;<super> 0030;;0;0;N;SUPERSCRIPT DIGIT ZERO;;;;
   <dbaron> we should probably be adding this new type to
            http://dev.w3.org/csswg/css3-lists/#numeric as well
   <fantasai> yeah
   <fantasai> http://dev.w3.org/csswg/css3-lists/#super-decimal
   ACTION: howcome fantasai to move super-decimal counter type from GCPM to Lists
   previous action is on howcome+fantasai
   RESOLVED: super-decimal counter type moves from GCPM to Lists
   <anne> fwiw "It's easier to ask forgiveness than it is to get permission"

   ACTION: chrisl to email plinss regarding the removal from CSS2.1 of
           'without reliable resolution information' from intrinsic
           dimensions definition in section 3.1
   <anne> (i.e. just make the change)
   (actually, we want Peter and HP's opinion)
   <anne> (Are we skipping css3-flexbox and Fonts?)
   <sylvaing> (text-replace: "css3-flexbox" "gcpm")
   <anne> heh
   RESOLVED: GCPM page-bleed renamed to bleed
   RESOLVED: keep page floats in the GCPM WD

   szilles: I am concerned about the clarity of the dependencies between
            our outstanding drafts
   szilles: while i'm happy to publish this as a working draft, I'm
            concerned on our ability to resolve and fix the issues and
            conflicts that may arise due to those dependencies
   howcome: should we rename this module ?
   chrisl: garbage collection placeholder module ?
   <lunch />

ScribeNick: fantasai
   <dbaron> http://dev.w3.org/csswg/css3-flexbox/
   dbaron: we resolved to publish this awhile ago
   but it took me awhile to get it into the right format
   dbaron: The editor token passed through a lot of revisions
   dbaron: The most recent one had a lot of examples removed
   dbaron: and also a lot of prose change
   dbaron: So somebody ought to evaluate those changes
   dbaron: but I also want to get published, because we've been discussing
           this a lot
   dbaron: There are other related proposals too, but I'd like to publish
           what we have, just to get a draft out
   Daniel: Any outstanding issues we should address before publishing?
   dbaron: The major issue is interaction with other layout proposals
   discussion of Ian's affiliation
   Daniel: We preserve the affiliation at the time of the work being done

   Alex: Can you give me a summary?
   dbaron: A flexible box has children laid out either horizontally or vertically
   dbaron: You have a property to choose which direction they go in
   dbaron: What you do with the extra space ...
   dbaron: In the dimension the children are laid out in, you can put
           the space at the beginning, at the end, center the children,
           or spread the space out
   dbaron: Then in the direction that's not the direction you're laying
           out, you can align the children to one side or the other
   dbaron: You can also assign flex to some or all of the children
   dbaron: flex controls where the space gets distributed
   dbaron: That I think is the part that is implemented in both Mozilla
           and Webkit
   dbaron: There's an unimplemented box-ordinal part
   dbaron: Which lets you say that flex goes to groups of elements in
   dbaron: I'm not actually sure of the details, and the spec is all we
           have to go on wrt what it was intended to be because nobody
           implemented it
   dbaron: A lot of the sizes by default, without flex, deal with
           intrinsic sizes
   dbaron: You can sometimes flex down instead of up
   dbaron: Because usually you use the preferred width, but you can
           flex things down to the minimum width
   dbaron: There's also another piece that wasn't implemented, box-lines,
           that allows boxes to lay out in multiple lines
   dbaron: The draft might be short on mathematical details for things
           like preferred and minimum widths...
   dbaron: and some of that is its interaction with other parts of the
           CSS box model
   Alex: This is already implemented?
   dbaron: yes

   Alex: You don't know how it interacts with other models?
   dbaron: To some extent, not very well.
   dbaron: The important use cases... the one's where you really care
           about interaction with other models
   dbaron: The main one is putting a chunk of block-and-inline layout
           inside a xul flexbox
   dbaron: the other is building some widget-like thing that goes inside HTML
   dbaron: In the first case, essentially what you're doing is most of
           the time the chunk of block-and-inline layout takes up all
           the available space
   dbaron: so you're not using its intrinsic width
   dbaron: The not-very-well understood case is when you're using the
           intrinsic width for this
   dbaron: but shrink-to-fit.... it doesn't work very well half the time
           in the cases where we use it,too ????
   dbaron: .... in floats, most of the time authors give a width
   Alex: The challenge is to define the behavior of what David refers to
         as not very well
   Alex: The scenarios that we weouldn't really care about for the
         purposes of this
   dbaron: I think some of that is just getting a definition so everybody does
           the same thing, nevermind trying to figure out something good to do
   Alex: So this is flex layout, but also float:right or tables.. are there
         issues with that?
   dbaron: This is triggered by an additional value of the display property.
           If you float it, it's a block, no longer a flexbox.

   Alex: this is used to create layout. Is this a complete layout model that
         is good for building user interfaces, or are there other pieces
         that you'd want?
   dbaron: This draft does not have the grids that we use in it
   dbaron: The firefox UI is built using the primitives here plus a grid
           that sort of works the same way, except it lays out its children
           in both dimensions instead of one
   anne says something about tables
   dbaron: Our grids are weird, because they're neither row-primary or
   dbaron: I don't know if it's something we'd be happy standardizing
   jdaggett: good weird or bad weird?
   dbaron: some of both
   dbaron: you can have a grid, give it a set of rows, then a set of columns
   anne: They overlap?
   dbaron: yes (?)
   Chris: It might be useful to give some background on what XUL uses
   Daniel: you can also say it's already implemented and this is a
           standardization of the model use by XUL

   anne: I still think it's confusing that Ian Hickson is listed as active
         editor and from Opera Software
   Daniel: We can write formerly of Opera Software, or list him as a
           Previous Editor
   people prefer Previous Editor

   Alex: I don't think it conflicts with grid-positioning. Overlaps, probably yes
   dbaron: I think there's a lot of overlap
   dbaron: There's been some discussion on www-style about ways to do flex
           in other things rather than just the box
   dbaron: For example, making margins flexible
   dbaron: Some people have been saying flex could be units.. there have
           been a lot of ideas thrown in here
   Daniel: From my experience building apps with XUL, what we have here
           is adequate
   fantasai: You'd have to use a lot of extra elements
   dbaron: Yeah, in XUL you typically use spacer elements
   jdaggett: Is there interest at Apple for doing more with this?
   dbaron: I think so. I think they use this in things like iPhone apps
   dbaron: and widget things that use WebKit
   Anne: Opera has interest in adding a third implementation, if it's
         better defined than now
   Anne: We haven't started because we think adding a third implementation
         of vagueness wouldn't be a good idea
   RESOLVED: Change Ian's editor's status and publish flexbox as FPWD

Received on Wednesday, 17 June 2009 07:48:20 UTC