Minutes: 2 May 2011 FXTF telcon


                   CSS-SVG Task Force Teleconference

02 May 2011


      [2] http://lists.w3.org/Archives/Public/public-fx/2011AprJun/0059.html

   See also: [3]IRC log

      [3] http://www.w3.org/2011/05/02-fx-irc


          Shepazu, hober, ed, +1.206.675.aaaa, anthony, rik,
          [Microsoft], +1.650.253.aabb, TabAtkins

          smfr, chris




     * [4]Topics
         1. [5]Filter Effects 1.0
         2. [6]'enable-background' deprecation
         3. [7]transform atribute vs. property
         4. [8]CSS/SVG Animation, FX 2d Transforms updates
     * [9]Summary of Action Items

   <trackbot> Date: 02 May 2011

   <scribe> ScribeNick: hober

Filter Effects 1.0

   shepazu: if someone defines a filter effect, but it doesn't exist or
   isn't supported, what should happen?


     [10] https://dvcs.w3.org/hg/FXTF/raw-file/tip/filters/publish/Filters.html#FilterFunction

   cabanier: if they don't exist they should be ignored

   ed: we apply the null filter in this case
   ... which makes elements transparent
   ... firefox, opera, webkit all make the element invisible

   shepazu: this is very unintuitive
   ... will follow up in email

   ed: it would be nice if the element remained visible

   shepazu: what if you have a set of filters
   ... you support a b and c, and don't support d
   ... in a filter chain a b d c, what is the input into filter c
   ... expects d to simply pass through the ouptput of b to c

   ed: in the svg case, you can have sub-parts that aren't supported;
   what should happen in such cases is hard
   ... pass-through sounds fine to me

   shepazu: also need to address this to clip paths
   ... if there's nothing to apply, nothing should happen

   <ed> feImage in firefox might be an example, supports raster images
   but not referencing other svg elements IIRC

   shepazu: does anyone have rationale to this behavior besides it's
   already how implementations behave?
   ... I tried to apply some filter effects to a video
   ... regardless of if it applies the effect I still want the video to
   be visible
   ... we should nip this in the bud now
   ... people won't use filters because of the very negative effect

'enable-background' deprecation

   cabanier: having access to the background is more about blending,
   should go in possible future blending spec

   <TabAtkins> ...dammit, I meant to be in the call. One sec.

   cabanier: if we allow enable-background in filters, we then have
   more than one way of expressing the same thing

   ed: describes example of wanting to blur a background while applying
   a fitler; background may be unusually large

   shepazu: filter within a filter with enable-background gets
   complicated really quickly
   ... we need to keep the filter spec simple

   ed: we still have to deal with backwards compat; there are some
   impls that do some of this
   ... but there isn't much content on the open web that contains this

   pdengler: which implementations support this?

   ed: batik, opera, both support some of it
   ... adobe plugin had support for it

   pdengler: would you deprecate it in opera?

   ed: enable-background is still used in the compositing spec

   shepazu: are you doing filters in IE? what happens now in IE9?

   pdengler: IE9 ignores it

   [questions about svg v. css compositing spec work]

   TabAtkins: is identical to what's being specced in svg

   pdengler: we're not changing svg filter spec

   shepazu: svg filter spec is becoming *a* filter spec

   pdengler: the importance is that we have a consistent model now, so
   that 2 or 5 years from now authors have consistent experience

   ed: there are some tests, but not many

   pdengler: if someone were to support enable-background when
   blending, would they break the future?

   ed: blending in terms of compositing or svg filters?
   ... don't know if you can get the same behavior between
   enable-background in compositing v. svg filters

   pdengler: there wouldn't be a collision

   cabanier: in a css compositing spec, should enable-background even
   be there
   ... if you have an animation, it's hard to implement that

   ed: can we refer to the compositing spec for the definition of
   ... background-image and background-alpha are the only things using
   enable-background property
   ... removing, we'd have to do something for existing content, not
   sure what

   <scribe> ACTION: ed to create test cases for background-image and
   background-alpha to see what current implementations do [recorded in

   <trackbot> Created ACTION-31 - Create test cases for
   background-image and background-alpha to see what current
   implementations do [on Erik Dahlström - due 2011-05-09].

transform atribute vs. property

   ed: cameron isn't here

CSS/SVG Animation, FX 2d Transforms updates

   anthony: next week i plan to make the fx 2d transforms updates

   pdengler: question about integrating transform attribute v. property

   shepazu: we've gone back and forth on that
   ... from an authoring point of view, keeping them separate is

   ed: cameron was objecting to it
   ... haven't seen a proposal that addresses all of the issues with
   making it a property; the dom issues, etc.
   ... might make it harder to promote other attributes to properties
   for animating with css animations
   ... we should see if there are more detailed comments from cameron
   on that
   ... we would have to change the fx transforms spec to reflect that
   ... any updates on the animation work in the css group?

   TabAtkins: nothing's changed
   ... dean's been working on a better js api for animations

   ed: that's not been checked in anywhere

   TabAtkins: dean will move that along once he's fleshed out his
   experimental impl

   pdengler: will there be an fx meeting in kyoto?

   [discussion of when the overlap day will be in kyoto]

   ed: we should know svg meeting dates by thursday
   ... the intention is to cover fx material in overlap mtg; not sure
   how many people doing the fx work will be there

Summary of Action Items

   [NEW] ACTION: ed to create test cases for background-image and
   background-alpha to see what current implementations do [recorded in

   [End of minutes]

Received on Tuesday, 3 May 2011 19:02:26 UTC