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

[CSSWG] Minutes Telecon 2013-07-31

From: fantasai <fantasai.lists@inkedblade.net>
Date: Wed, 31 Jul 2013 23:46:56 -0700
Message-ID: <51FA0460.4000105@inkedblade.net>
To: "www-style@w3.org" <www-style@w3.org>
   - Discussed proposal to allow justification in combination with
     explicit letter-spacing
     Seem to have consensus around Part I of proposal, except for Bert

   - Plan to tie impact of 'image-resolution's 'snap' keyword to
     layout-affecting zoom; transforms and scaling-zoom would not have
     an effect. Tab to propose text describing the two different kinds
     of zoom in either Media Queries or CSS Device Adaption.

   - RESOLVED: two X/Y values for image-resolution, to allow explicit
               values to match from-image in capabilities

   - RESOLVED: Clarify spec that CSS units (not physical units) are
               used for resolutions taken from image data

   - RESOLVED: FPWD Matrix, with naming issues noted

   - RESOLVED: box-fixup on internal table elements before flex item
               determination, unless further problems raised during LC

   - WG members should register participation in Paris F2F on wiki

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

   Glenn Adams
   Tab Atkins (late)
   Shezan Baig
   David Baron
   Bert Bos
   Tantek «elik
   Elika Etemad
   Simon Fraser
   Sylvain Galineau
   Daniel Glazman
   Rebecca Hauck
   Koji Ishii
   Dael Jackson
   John Jansen
   Brian Kardell
   Philippe Le Hťgaret
   Peter Linss
   Chris Palmer
   Florian Rivoal
   Simon Sapin
   Dirk Schulze
   Alan Stearns
   Leif Arne Storset
   Lea Verou
   Steve Zilles

<RRSAgent> logging to http://www.w3.org/2013/07/31-css-irc
Agenda: http://lists.w3.org/Archives/Public/www-style/2013Jul/0685.html
Scribe: SimonSapin

CSS3 Text

   fantasai: on the ML, edits to justification section
   <glazou> see also https://lists.w3.org/Archives/Member/w3c-css-wg/2013JulSep/0092.html
   fantasai: waiting for SteveZ and jdagget to review and approve
   fantasai: if no comment, next issue
   SteveZ: jdaggett wants to continue discussion on the ML

   [discussing what is the 2nd issue]
   fantasai: proposal: letter-spacing allows justification with space
             between characters when set to a length
   fantasai: consistent with word-spacing, impls., some content depends
             on this
   SteveZ: agree to allow some level of justification if letter-spacing
           is used
   SteveZ: not happy with 'fixed' as a way to turn it of
   SteveZ: letter spacing variation are very small (few %)
   SteveZ: fixed doesnít correspond to what people find useful: min,
           mix variation
   fantasai: one issue at a time
   fantasai: 1 whether letter-spacing: length suppresses justification
   fantasai: 2. do we have a way to turn this kind of justification off
   SteveZ: is 2. a way to control it?
   SteveZ: control is more important
   fantasai: already resolved to not allow min/max spacing at this level
             of the spec
   SteveZ: if you canít control it, you shouldnít allow it
   fantasai: then content breaks, impls need to change, and this is
             inconsistent with word-spacing
   SteveZ: lets put min/max back on the table
   fantasai: not going to CR if you say we need this, and jdagget says
             we canít have this

   <plinss> ack Bert
   <Zakim> Bert, you wanted to suggest the question is the wrong way around:
           we need a way to turn flexible letter spacing *on* (not off)
   Bert: the way to turn automatic letter-spacing on could be to use the
         'distribute' keyword
   Bert: maybe not necessary to have control on the limits, but just say
         there is a limit or no limit on expansion
   Bert: maybe not further than twice the normal size is good enough
   Bert: new keyword on text-justify, suggested on the ML. 'unlimited'
   SteveZ: this is what fixed does
   SteveZ: the spec does not specify a limit, but gotta be reasonable
   SteveZ: kinds of limits I was coming across are +/- 5%, much smaller
           than half

   fantasai: letter-spacing also applies between CJK characters, in this
              case you do not want to limit
   SteveZ: CJK task force has a huge table of cases, itís not uniform at
           all. Unclear that this works for CJK
   fantasai: auto justification algo is undefined, UAs encouraged to do
             the right thing
   fantasai: don't want to have strong limits on what UAs can do: 2.1 says
             "can not add space between characters for justification"
   stearns: 'fixed' keyword is not about not allowing variable expansion,
            it forbids expansion at all

   <stearns> thought that 'distribute' is what Bert is suggesting as
   <Bert> (To stearns: yes, and I suggest redfining 'distribute' as including
           an implicit limit, and 'unlimited' is what 'distribute' does now.)
   <stearns> Bert: I'd rather leave 'distribute' as is

   fantasai: goal is to allow CJK to justify correct, so need to lift this limit
   fantasai: also not break content and impls
   fantasai: in order to get the previous spec behavior: add the 'fixed'
             keyword, if thatís what you want
   fantasai: impls will have to add it, but does not break content as its opt-in
   fantasai: can add further controls in the future

   plinss: anybody implemented previous spec behavior?
   fantasai: not that I know of
   Bert: I've been relying on it. letter-spacing: 0
   fantasai: people who donít read specs donít do that, because currently
             it has no effect.
   fantasai: Implementations right now don't do justification with spacing
             between latin letters.
   Bert: content is there for what the spec says, not for future impls
   florian: if nobody has implemented it, [???]
   florian: I think fantasaiís way forward is more managable
   plinss: continue discussion on email?
   <stearns> +1 to fantasai's current wording
   <florian> If nobody has implemented it, I suspect not many people have
             written stylesheet that conform to the spec in a way that
             breaks on current implementation, so while it is unfortunate
             to contradict ourselves, it still sounds like a less painful

   SteveZ: I think there is some agreement to allow letter-spacing to
           participate in justification
   SteveZ: weíre struggling with how to do that with existing impls/spec/content
   SteveZ: even if we add min/max, you have to turn those one which doesnít
           work with existing content
   SteveZ: unless we have defaults like +/- 5%
   fantasai: thatís too small for CJK
   stearns: leave impls. to choose limints
   stearns: can have controls for the limits later
   stearns: in favor of fantasaiís proposal now
   SteveZ: I may be ok with that

   SteveZ: what happens if you say fixed and specify a range
   fantasai: you can't
   fantasai: Basically, if we add min/max controls in the future,
             'fixed' will be a shorthand to specify 3 identical values
   fantasai: never able to combine it with a range
   SteveZ: to do this correctly you need a table for CJK
   SteveZ: table = range of values depending on context
   SteveZ: also priorities between adjustments
   SteveZ: more than %age, more complex in the Japanase Layout Task Force
   fantasai: letís not design that solution at this level
   SteveZ: concern with 'fixed' is that it restricts this solution
   SteveZ: letter-spacing was originally designed for latin

   fantasai: want UAs do the right thing for justification by default,
             though it may take awhile before we quite get there.
   fantasai: fine tuning of this is not something we should do now, if at all
   SteveZ: OK with that, I just donít like 'fixed'
   SteveZ: can we live without it?
   fantasai: Iím ok with that
   stearns: one of the use case for 'fixed' is German text, disable letter
            spacing for justification to avoid confusion with emphasis
   SteveZ: ???
   SteveZ: when we see problems, we can engineer the right solution
   SteveZ: 'fixed' seems to be not terribly helpful
   plinss: consensus?
   fantasai: I think allowing justification for 'letter-spacing: <length>'
             and not having fixed is what jdaggett originally wanted,
             so I think we should just resolve on not having 'fixed' and
             he can object if he wants.

   Bert: "There are newspapers that do that - more than 5%"
   Bert: what if you do want letter spacing for justification?
   fantasai: undefined for now
   <tantek> perhaps post screenshots of newspapers that do this?
   SteveZ: we say UAs should "do the right thing"
   SteveZ: we need to experiment to see what values/controls make sense
   Bert: By default I want that limit at 0 or 5%
   Bert: donít want to leave it completely open. Impls will do letter-spacing
         without any limit and we wonít be able to get rid of it anymore
   Bert: "Would like some way to say, if you use this keyword, then
         you may use more than 5%"
   plinss: the default is to whatever you think is right, 'auto' keyword
   SteveZ: when mixing Latin and CJK, no one single number gives a good answer

   fantasai: proposed resolution: accept part one of the proposal
   Bert: I do not want to allow that between alphabetic letters
   fantasai: you have to allow it for CJK, and need to allow more than 5%
   SteveZ: bertís proposal is to only relax when you say 'distribute'
   Bert: 'auto' means letter-spacing is honored
   Bert: I have content with letter-spacing:0 because I donít want expansion
   SteveZ: existing content that depends on the non-spec behavior of
           existing impl
   fantasai: existing content has letter-spacing:0 and expect expansion
   Bert: thatís not what the spec says, we donít have to deal with that
   ???: yes we do
   plinss: letís move on

   fantasai: my understanding is: consensus except for Bert
   fantasai: discuss with Bert on the ML?
   SteveZ: would be helpful to document what existing content would break
   fantasai: CJK content (no spaces) with 'letter-spacing: 0' that expects

   Topic: text-align
   Bert: problem that fantasai mentioned is the cascading problem
   Bert: not sure thatís the same
   <dbaron> (fantasai seems to have dropped off the call)

Conditional Rules

   plinss: where are we? Moving the spec forward
   dbaron: I donít really know
   plinss: can we look into it and come back to it next week?
   dbaron: T&A is higher priority

Scribe: fantasai

image-resolution: snap

   SimonSapin: wrt snap keyword of image-resolution
   SimonSapin: It's not really well-defined in CSS what the resolution is
   SimonSapin: esp. wrt zoom and transforms
   SimonSapin: Consensus on ML seems to be that transforms don't affect snap
   SimonSapin: Zoom that changes size of viewport should affect snap,
               but purely "optical" zoom should not
   TabAtkins: Need some place that actually defines concept of viewport-zoom
              vs. other zoom
   TabAtkins: This distinction also affects device-pixel-ratio etc.
   TabAtkins: The things that 'snap' responds to are same as canvas
   TabAtkins: Dunno where to define
   fantasai: I think MQ is a good place to define this
   SimonSapin: What about device-adapt spec?
   fantasai: That might be ok, too. What is the status of that anyway?
   plinss: No WD since 2011
   TabAtkins: should poke Opera
   <sgalineau> is the editor still at Opera?
   <oyvind> device-adapt? yes
   florian: yes, the editor is still at Opera
   ACTION TabAtkins: Define zooming, 2 types, for insertion into either MQ or device-adapt
   <trackbot> Created ACTION-572
   TabAtkins: visual zoom vs. layout zoom
   fantasai: Define snap to respond only to layout zoom
   <dbaron> I would *not* use the terms "visual zoom" and "layout zoom"
            that TabAtkins suggested
   <dbaron> The distinction really has to do with whether there's one
            viewport or two.
   <TabAtkins> Suggestions welcome, dbaron. ^_^

   SimonSapin: 2 more issues...
   SimonSapin: Two values for horizontal and vertical resolution
   fantasai: Think that's out of scope for L3
   SimonSapin: But when you have from-image, some images can have 2 values
   SimonSapin: So CSS should also be able to handle that
   florian: Given from-image is in this level, maybe do it in this level
   fantasai: Could just allow it via from-image
   fantasai: Ordering of dimensions should be same as border-spacing,
   fantasai: Note it's physical
   TabAtkins: Well, logical in relation to the image
   SimonSapin: Will interact with image-orientation
   fantasai: yep
   SimonSapin: Move to ML for details?
   fantasai: Sounds reasonable. Maybe draft up text and bring back to WG?
   fantasai: Anyone else on this topic?
   RESOLVED: two X/Y values for image-resolution

   SimonSapin: Units for image-resolution from-image
   SimonSapin: from-image metadata, e.g. png spec has number of image
               pixels per cm or whatever
   SimonSapin: Do we interpret that as CSS units rather than physical units?
   fantasai: Yes
   SimonSapin: Clarify in spec
   RESOLVED: Clarify spec that CSS units are used for from-image resolution
             as well as CSS-explicit resolutions

Matrix API

   <krit> http://dev.w3.org/fxtf/matrix/
   krit: We wanted to have an interface that can handle 3D as well
   krit: Could have Matrix interface used by SVG and CSS together
   krit: Hopefully CSSOM will eventually expose the matrix interface
   krit: So wanted a joint specification
   krit: would like to ask for feedbac, fpwd
   * glazou supports FPWD for CSSMatrix interface
   krit: Already asked for review 3-4 weeks ago, no feedback

   plh: You name the interface CSSMatrix
   plh: Do you imply it can only be used by CSS?
   krit: Was called Matrix before, but not happy for WebGL people
   krit: Not useful for them
   krit: Asked us to use a more specific name
   krit: since used for CSS Transforms, called it CSSMatrix
   <dbaron> I'm not convinced by the argument that it should have
            a CSS prefix
   <tantek> dbaron++
   * fantasai too

   Bert: If we make this SVGMatrix, maybe SVGWG can take care of
         publishing? ;)
   glazou: Is name of interface a blocker for FPWD?
   <tantek> I suggest we go FPWD without prefix
   dbaron: I don't think it is, but should note the issue.
   <tantek> OH: "Ö then wait for last call to change the name :)"

   <sgalineau> TransformMatrix?
   * fantasai likes it!
   smfr notes that there's also CSSPoint interface
   krit: Also have a DOMPoint interface
   krit: Think I added an issue... it's under discussion.

   fantasai: So, note the issues, publish FPWD?
   RESOLVED: FPWD Matrix, with naming issues noted
   * tantek is pleased to see the Matrix make progress.


   fantasai: I guess we discussed converting table-cells to flex items
   fantasai: Do you have @supports yet? I think I would be uncomfortable
             not having good fallback from flex to tables if we don't
             have good support for @supports

   Rossen: min-size?
   TabAtkins: Different issue.
   TabAtkins: Read & comment on thread

   fantasai: Think we can go with box-fixup clarification
   fantasai: and revisit during LC if necessary
   RESOLVED: box-fixup on internal table elements before flex item

<dbaron> If you're planning to come (or might come, please list probability)
          to the Paris F2F, please add yourself to
Meeting closed.
Received on Thursday, 1 August 2013 10:09:57 UTC

This archive was generated by hypermail 2.4.0 : Friday, 25 March 2022 10:08:32 UTC