CSS-SVG Task Force Teleconference

11 Oct 2010


See also: IRC log


+1.408.636.aaaa, smfr, ed, ChrisL, Shepazu, anthony


<trackbot> Date: 11 October 2010

<scribe> scribenick: ChrisL

CSS SVG transforms update

<ed> http://lists.w3.org/Archives/Public/public-fx/2010OctDec/0000.html

anthony: see http://lists.w3.org/Archives/Public/public-fx/2010OctDec/0000.html

and http://dev.w3.org/Graphics-FX/modules/2D-transforms/spec/2DTransforms.html

anthony: sumary at first url, spec at second one
... some email from dr Olaf, pointed out some things

<anthony> http://lists.w3.org/Archives/Public/public-fx/2010OctDec/0001.html

anthony: mainly concerns related to changes from SVG

ds: wish there was a summary

anthony: I made one
... changed todo list, added some items
... as follows

<anthony> - Address any issues with transform-origin

<anthony> - IDL

<anthony> - Address if the properties are independent of transform attribute used in SVG (1.1, 1.2)

<anthony> - Address if the transform properties are available as presentation attributes

<anthony> - Investigate and address backwards incompatibilities (rotate, origin on which transform is applied)

<anthony> - Address how to determine the difference between original SVG attribute and new presentation attribute (example)

<anthony> - Address if 'transform-origin' property has any influence on appearence of old SVG transform attribute

<anthony> - Address issue for objects with no bounding box (e.g. line)

smfr: problem with assuming identity is the same as none
... in CSS identity doews something different, establishes a stacking context
... these need to be different values. initial value of none, bt identity as a defined value to start or end from
... transform-origin initial value in SVG?
... 50% 50%?

anthony: odd for it to do differnt things in css and svg

smfr: how so?
... for pointer events we have a different initial value for svg and html
... could do the same here

shepazu: why have it different?

smfr: fill and stroke dont translate to html

ChrisL: yes that is a perfectly reasonable for pointer-events

shepazu: do we have to have it different?

ChrisL: if svg already has them and css does not but has stacking contexts then the initial value needs to be different

ed: could see how many would brewak with a 50% 50% transform origin
... probably a lot

shepazu: can we make it so that both behaviors work in svg somehow

anthony: given that transform-origin is not previously part of svg, its new so should not be too much of an issue
... but for transform property it may break

ChrisL: if its a property then it always has a value. even in older content that does not specify it explicitly

ed: would auto be ok as an initial value

ChrisL: our initial value is 0 0 in the local coordinate system. that can't be expressed with percent syntax which is relative to the bbox

ed: transform-oriin accepts coordinates?

smfr: well, it accepts lengths
... percent is percent of the border box

shepazu: how to say 50% of the document?

smfr: you can't

ChrisL: so the initial value for svg is 0 0 which is relative to the local coordinate system

<ed> ED: so if we go for "0 0" as the initial value for svg, you would just have to specify transform-origin:center to get the 50%,50% behaviour, not that much work really

smfr: in svg, bbox is pre-transform?

ChrisL: yes

shepazu: can you have an api to get the bbox?

smfr: yes there is one that gets the bbox of the transformed element. awkward when it maps to screen space
... existing apis use bbox
... have also had request for api to map local to global coordinate system

shepazu; dom3 has a request to hace such an api, jwatt made that request

scribe: anticipate html/css will need that

shepazu (lloks for link)

<shepazu> http://lists.w3.org/Archives/Public/www-dom/2010OctDec/0013.html

smfr: resisted giving the cuulative transform, because with 3d transforms flattening can occur, to put a plane in another 3d space so its hard to describe the whole thing

anthony: yes you would need a whole transform list
... happy to go with the group. initial value for transform-origin, would be nice to be the center, and then somehow ...

ChrisL: that would change behaviour for existing content in new and old implementations. it would break all existing content that uses transforms

ed: resolution?

resolution: initial value of transform-origin is auto, which in svg means 0 0 and in html/css means 50% 50%

smfr; so using css to apply this to svg would have the traditional 0,0 origin point?

ChrisL: yes

<scribe> ACTION: anthony to add the new initial value of transform-origin to the spec [recorded in http://www.w3.org/2010/10/11-fx-minutes.html#action01]

<trackbot> Created ACTION-17 - Add the new initial value of transform-origin to the spec [on Anthony Grasso - due 2010-10-18].

Premultiplied gradients in CSS3




ed: css3 seems to want to have premultipied gradients, different from canvas and svg

smfr: otherwise gradient from red to transparent has grey in it

ChrisL; that is because transparent is rgba(0000) which is transparent black

scribe: if you want a fade to red, fade to transparent red not transparent black

ed: people would want the css3 gradient systax with canvas and svg. they will be surprised

smfr: could we add a value to color interpolation to say its premultiplied

ChrisL: also in premultiplied, lower resolution for colors as the transparency goes up

smfr: so options are to pushback on premultiplied, or to add a value to allow controlling the interpolation

ed: not that difficult to add

shepazu: more properties

ed: no, existing property with one new value

ChrisL: concerned about hue shifts with fades of pastel colors

<ed> http://www.w3.org/TR/SVG11/painting.html#ColorInterpolationProperty

smfr: we use premultipiied already for gradients in safari and firefox, certainly for transitions we use premultiplied
... underlying lirary on mac does not give us a choice
... chris, please post an example

<scribe> ACTION: chris post examples of reduced resolution with premultiplied colors [recorded in http://www.w3.org/2010/10/11-fx-minutes.html#action02]

<trackbot> Created ACTION-18 - Post examples of reduced resolution with premultiplied colors [on Chris Lilley - due 2010-10-18].

smfr: we use this for color transitions and for css3 gradients. that spec is not finished though

'writing-mode' values across CSS and SVG

ed: hoped to have elika here for this one
... http://www.w3.org/Graphics/fx/track/issues/3



all implementations do different things with writing-mode. want to alighn with svg but there is inconsistet implementation behaviour

scribe: wanted to see if spec defines correct behaviour

ChrisL: do we have a table of results for different implementations on that test?

ed: no. put this topic off until elika is here
... batik does 123 on second line in orange, in differnt way to opera. not well tested code in opera, not much used
... not that well defined in spec either.

ChrisL: good topic for tpac, needs a whiteboard

anthony: hope to be there

ed: will be there too
... good to define transforms more at tpac too


<scribe> agenda: http://lists.w3.org/Archives/Public/public-fx/2010OctDec/0008.html

Summary of Action Items

[NEW] ACTION: anthony to add the new initial value of transform-origin to the spec [recorded in http://www.w3.org/2010/10/11-fx-minutes.html#action01]
[NEW] ACTION: chris post examples of reduced resolution with premultiplied colors [recorded in http://www.w3.org/2010/10/11-fx-minutes.html#action02]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2010/10/11 20:57:23 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.135  of Date: 2009/03/02 03:52:20  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/who/how/
Succeeded: s/valie/value/
Succeeded: s/value/value of transform-origin/
Succeeded: s/does the right thing/would have the traditional 0,0 origin point/
Succeeded: s/soec/spec/
Found ScribeNick: ChrisL
Inferring Scribes: ChrisL
Default Present: +1.408.636.aaaa, smfr, ed, ChrisL, Shepazu, anthony
Present: +1.408.636.aaaa smfr ed ChrisL Shepazu anthony
Agenda: http://lists.w3.org/Archives/Public/public-fx/2010OctDec/0008.html
Found Date: 11 Oct 2010
Guessing minutes URL: http://www.w3.org/2010/10/11-fx-minutes.html
People with action items: anthony chris

[End of scribe.perl diagnostic output]