Re: Agenda request: Presentation attributes in animation sandwich model

Hi Dirk,

I'm just sending a quick note since I won't be able to join the telcon. 
Initially I was a bit concerned about prioritising animations based on 
syntax. Taking a holistic view of animations on the Web, I wondered if 
there is a more natural way to prioritise animations rather than syntax. 
Animation start time, which is what SVG uses (plus a few extra 
conditions), is just one example.

Eventually in CSS we will probably want multiple animations on the same 
property (e.g. transform) to add together which will mean we'll need a 
model for prioritising the respective animations within CSS anyway. So, 
I thought, why not have one rule that covers all animations regardless 
of syntax. This would seem to make a lot of sense if SVG and CSS 
animations overlap more and more and become more and more 
interchangeable in the future (e.g. invoking an SVG animation via CSS' 
animation-name property).

Unfortunately, coming up with such rules takes time and I understand 
there is a need to have guidelines now (and if we have those guidelines, 
authors will start to rely on them and we won't be able to change them). 
So I was a little concerned about painting ourselves into a corner here.

Having thought about this a little more (and having been persuaded by 
others), I think we can probably get away with the syntax divide. 
Ultimately the rule might look something like "animations applied 
through the style system sit on top of those applied directly to the 
elements." That would still allow a lot of the future situations I have 
in mind (e.g. CSS animations adding to SVG, SVG animations being applied 
through CSS etc.) Even if we were to phrase it as "CSS beats SVG" we 
could make it more generic later.

Also, even with such a rule, I've been persuaded that it will still be 
possible to add more fine-grained control later so that we could have, 
for example, an SVG animation adding on top of a CSS one.


I'd be interested in hearing the discussion on testing too. Mozilla have 
been using the strategy Nikolas outlined for a few years now and it 
works pretty well. I'd be particularly interested to know how this 
framework is being applied to CSS animations/transitions (since it seems 
to be based around having a seekable timeline).

Thanks,

Brian

(2012/02/25 5:42), Dirk Schulze wrote:
> Hi SVG WG,
>
> I have an agenda request with the topic "Presentation attributes in animation sandwich model" for the next meeting.
>
> This comes up after reviewing the CSS3 Animation specification[1]. The current specification does not describe how CSS3 Transition or SMIL animation of presentation attributes contribute to the animation sandwich model.
>
> We already have a resolution how to solve this issue for the 'transform' attribute (and all new presentation attributes with SVG DOM interfaces)[2]. SVG Animation will take the value of the intrinsic styles, when no CSS animation is running, and takes the contributed results from CSS3 animation if an animation is running. The graphic attached to this mail should help to understand this model.
>
> The problem does not only occur on 'transform', but all existing presentation attributes. SMIL 3.0 Animation[3] is outdated and does not consider CSS Animations nor CSS Transitions. It would make most sense, when CSS3 Animation gets updated to consider all three types of animations.
>
> I would like to see the resolution for some presentation attributes get extended to all presentation attributes, so that we the editors of CSS3 Animations have the blessing of the SVG WG. According to Daniel Holbert (a SVG contributor at Mozilla), this is already done that way in Gecko.
>
> Greetings,
> Dirk
>
> CC'ing www-svg to address more people.
>
> [1] http://dev.w3.org/csswg/css3-animations/
> [2] http://www.w3.org/Graphics/fx/wiki/SVG_attribute_to_presentation_attribute
> [3] http://www.w3.org/TR/SMIL3/smil-animation.html#animationNS-AnimationSandwichModel
>

Received on Tuesday, 28 February 2012 06:23:28 UTC