W3C home > Mailing lists > Public > www-svg@w3.org > January 2012

minutes, 19 January 2012 SVG WG telcon

From: Cameron McCormack <cam@mcc.id.au>
Date: Mon, 23 Jan 2012 10:03:06 +1100
Message-ID: <4F1C95AA.4010509@mcc.id.au>
To: SVG public list <www-svg@w3.org>
Hello, please find below the minutes from last week's SVG WG telcon. 
Although the minutes do not state it, Vincent also sent his regrets.



       [1] http://www.w3.org/

                                - DRAFT -

                    SVG Working Group Teleconference

19 Jan 2012



    See also: [3]IRC log

       [3] http://www.w3.org/2012/01/19-svg-irc


           ed, heycam, krit, Tav, ChrisL, +1.732.216.aabb

           Rik, Cyril




      * [4]Topics
          1. [5]svg in border- and background-image
          2. [6]svg2 requirements
          3. [7]snapshot time
          4. [8]the 'timelineBegin' attribute
          5. [9]the 'requiredFonts' and 'requiredFormats' conditional
             processing attributes
          6. [10]the Progressive Rendering algorithm
          7. [11]the <prefetch> element
      * [12]Summary of Action Items

    <trackbot> Date: 19 January 2012

    <scribe> scribenick: ChrisL

svg in border- and background-image

    ed: raised in css images spec a long time ago


      [13] http://lists.w3.org/Archives/Public/www-style/2011Dec/0427.html

    ed: asking what to do when border-image references an svg
    image.slices into 9,not clear how it should work - split beforeor
    after computing aspect ration
    ... elika says if we want it to work a particular way we should
    define it, not css

    heycam: does css spec give us some hooks?

    ed: not sure

    <ed> [14]http://dev.w3.org/csswg/css3-images/

      [14] http://dev.w3.org/csswg/css3-images/

    heycam: section 3 in there

    ed: also section 5.1

    <ed> [15]http://dev.w3.org/csswg/css3-images/#sizing

      [15] http://dev.w3.org/csswg/css3-images/#sizing

    heycam: assume the way its defined is that you put the box for the
    outer svg element for the size before slicing up and nothing special
    to be done for aspect ration

    ed: so you then stretch the bitmap

    heycam: why

    ed: where you renderthe border in a different aspect ratio

    ChrisL: prefer to see this as clipping regions rather than
    re-scaling a raster result

    heycam; like if you had an svg element in an htmldoc and set the
    width and height properties

    <krit> krit in th channel yes

    krit: is that on the border width of the element that uses svg as
    the border image?

    heycam: maybe something can be done with intrinsiic aspect ratio?

    krit: we have default values for width and height

    heycam: the 300 150?

    <krit> :)

    heycam: whitespace on top and bottom makes it less useful. put PAR
    ... if you use a raster image, the middle parts get scaled?

    ChrisL: you can scale them or replicate them

    <krit> [16]http://www.css3.info/preview/border-image/

      [16] http://www.css3.info/preview/border-image/

    Tav: need an example

    ChrisL: a lot of stretch, which is why its better to do as a clip
    and render atthe exact size needed

    <ed> [17]http://www.w3.org/TR/css3-background/

      [17] http://www.w3.org/TR/css3-background/

    heycam: sounds like we know what we want to put in the spec

    ed: is there a test suite

    ChrisL: a bit

    ed: and none for this particular case? can take action to make some

    <scribe> ACTION: erik to make css3 background and border tests
    [recorded in

    <trackbot> Created ACTION-3228 - Make css3 background and border
    tests [on Erik Dahlström - due 2012-01-26].

    heycam: suggested text that eleika pointed to in her second mail


      [19] http://lists.w3.org/Archives/Public/www-style/2012Jan/0809.html

    krit: from the second email we would not be able to use repeat,
    looks the same as stretch

    heycam; seems like its done after the

    ChrisL: need to work out an integer number of repeats, scale
    slightly so theywork out to an integer multiple

    krit: yes

    heycam: if you have an explicit w&h on svg,can use that to do the
    ... but if only an intrinsic aspect ratio

    Tav: nothing to divide by

    heycam: you dont have the intrinsic width

    ChrisL: aim is to stretch the replicated part as little as possible

    heycam: fit each of the 8 sections separately ... sounds sensible

    ChrisL: we can define this and also give authoring guidelines to use
    it well

    heycam; so weshould propose some wording back

    ed: wanted to hear what group thought of responses so far. want what
    is best for authors
    ... so lets go for the separate sections
    ... getting a feel for what is most natural

    heycam: good to work the same as for raster - consistency

    ChrisL: while still gettingthe benefit of vector

    heycam: yes
    ... someone should look into the repeating while preserving aspect
    ratio before repeating

    <scribe> ACTION: chris to look into the repeating while preserving
    aspect ratio before repeating [recorded in

    <trackbot> Created ACTION-3229 - Look into the repeating while
    preserving aspect ratio before repeating [on Chris Lilley - due

svg2 requirements



    heycam: we are on the tiny 1.2ones - i will edit as we go

snapshot time

    heycam: i like that
    ... as a way to render animated documents for a static view
    ... easier to specify a particular time

    ChrisL: it is good, but only for implementations that do animation
    and need a paster image

    krit: if you have multipleanimations?

    ChrisL: yes, its on the whole document

    krit: and events
    ... events that are generated?

    ChrisL: yes, you run the whole animation intil the clock is the
    snapshot time, then you render that state

    heycam: in batik i did a set current time
    ... so you get all the syncbase stuff. not sure if hyperlinking
    dispatches the start and end events
    ... better to do like that rather thsn have two ways to seek through
    the document

    ChrisL: suggest we accept the requirement and tighten up the spec on
    how you seek

    ed: spec does not say the snapshotTime image must be rendered for
    anything in the svg spec itself, it's suggested it may be used for
    still images, but doesn't require it to be used by the UA

    heycam: printingis a case, for html when would it

    ChrisL: (the use case was a set of animated icons, whose animation
    was disabled, but re-enabled on user interaction like mouseover) :)

    heycam: the snapshot time could be the base values before you run
    the animation

    ChrisL: need to stop it being displayed at the start of the
    animation though

    heycam: happy to bring this over - would still be metadata

    ChrisL: would be good to add a fragment like the view ones, to say
    render at the snapshot time

    ed: strange that its optional on 1.2T - we could tighten it up

    resolution: svg2 will add snapshot time for animated documents and
    may specify how to get at the rendered image

the 'timelineBegin' attribute





    heycam; wanted to get brian's input on the nextr few animation ones

    <ed> here's an example of timelineBegin:

      [24] http://dahlstr/



    heycam: ties into alignment with css animations, which do not wait
    for document load
    ... want to mail brian and ask before we decide on these

    ChrisL: makes sense

    ed: useful functionality, waiting for document load is poor for
    large files

    <scribe> ACTION: cameron to mail brian about thesvgt 1.2 animation
    items [recorded in

    <trackbot> Created ACTION-3230 - Mail brian about thesvgt 1.2
    animation items [on Cameron McCormack - due 2012-01-26].



the 'requiredFonts' and 'requiredFormats' conditional processing

    ChrisL: formats was to allow for audio and video where there is
    variable support

    heycam: html5 does it differently without a switch
    ... is it useful for content outside of video

    ed: for raster image formats and other special image formats; also
    for audio

    heycam; should use same mechanism as video

    ed: not so useful for image formats

    heycam; handling fallback for raster images is not well specified,
    but how often do people use other formats

    ChrisL: bitonal tiff is common for cad

    krit: so on video and audio, html5 does not use requiredFormats

    heycam; right you use the track and the first supported one is used

    scribe: you cant have other svg displayed conditional on the format
    support tough

    heycam: media types do not specify the codec
    ... what about codec registries

    ChrisL; yes the intent is to add a parameter to the media type,to
    specify the codec

    heycam; registry is for avi and wav containers

    scribe: not clear its up to date

    heycam: neutral on requiredFormats in general

    krit; we don't need requiredFormats because html5 video and audio
    don't use it

    ed: with downloadable fonts, requiredFonts is less useful

    ChrisL: does add value if the font is 404, and it allows tro depend
    on a set of fonts being available. css does not have that although
    it has been requested

    heycam; agree it is not otherwise available

    ed: opera implements them both, not seen that much use. ok either

    ChrisL: with dl fonts its actually more useful to be able to do
    conditional based on a set of fonts

    heycam; reqFonts is simpler than reqFornmats (no codec worries) so
    more in favor

    scribe: still unsure about the switch-heavy approach in general

    krit: neutral on requiredFonts

    <ed> you can use the required* attributes without switch too

    <ed> even if it's more useful with probably :)

    heycam: accept requirement for requiredFonts

    resolution: SVG2 wil adopt the requiredFonts attribute
    ... SVG2 will not adopt the requiredFormats attribute

the Progressive Rendering algorithm



    ChrisL: this is good clarification

    heycam; agree we should be more precise than with 1.1. Need to tie
    in with HTML5 document processing.

    scribe: start with the text here and modify it wrt html5

    resolution: SVG2 will add more precise text on progressive
    rendering, drawing on svgt 1.2 and html5

the <prefetch> element




      [30] http://www.w3.org/TR/SVGTiny12/struct.html#PrefetchElement

    ChrisL: its for explicit prefetch rather than relying on @w or @h

    heycam; in scripting this can be done with new image - but this
    works for non scripted

    scribe: has a bunch of attrs that are relevant for media, like how
    many seconds to prefetch and how much bandwidth to use(!)
    ... not sure about the media-specific things

    ChrisL: why

    heycam; conflict with html5, and not sure thay are super useful

    krit; spec says ua can ignore the prefetch hint

    heycam: can have an image with visibility: hidden which causes
    prefetch - but would not want the audio to start

    ed: opera did not implement this for media, just for images

    ChrisL: remember we resolved to have a audio voume property so
    itcould be prefetched and muted

    heycam: since we can prefetch anyway, not sure how useful the
    optional prefetch hint is
    ... first para says svg 1.1does not define when referenced resources
    start downloading. we should define thst, to make vis: hidden
    prefetching reliable
    ... then we would not need prefetch element

    krit: agree

    ed: not that useful

    ChrisL; ok

    resolution: svg2 will not have the prefetch element but will clarify
    when resources start doanloading

    <ed> i would like us to state whether resources start downloading as
    soon as you set the href, even if it's not appended to the document

    <ed> similar to Image in html, e.g var img = new Image(); img.src =

Summary of Action Items

    [NEW] ACTION: cameron to mail brian about thesvgt 1.2 animation
    items [recorded in
    [NEW] ACTION: chris to look into the repeating while preserving
    aspect ratio before repeating [recorded in
    [NEW] ACTION: erik to make css3 background and border tests
    [recorded in

    [End of minutes]
Received on Sunday, 22 January 2012 23:03:51 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:54:33 UTC