Re: SVG animations without SMIL

On Jun 2, 2015, at 8:19 PM, Philip Rogers <<>> wrote:

Can 'd' (or some solution for paths) be made a priority for SVG2? The reality is that it's not possible to create an SVG path animation that works in major browsers today and you have large projects shipping crippled SVG animations to major browsers. My impression is that folks were holding out for a SMIL implementation in all browsers but implementations seem to be heading in the opposite direction.

SVG2 is the foundation of a more agile approach that the WG is going. New version shall be published earlier but have less new features. I don't see a reason to rush in a d property at this state. We need a clean spec first to improve the spec quicker later.

That said, concrete suggestions for d are welcome. There is nothing that should stop the discussion about it. You can start by commenting on some of the mentioned issues with your thoughts.


On Tue, Jun 2, 2015 at 10:44 AM, Dirk Schulze <<>> wrote:

On Jun 2, 2015, at 7:29 PM, Philip Rogers <<>> wrote:


Animating SVG content without SMIL can be difficult and I'd like kick off a discussion on how to improve it. This recently came up because IE does not have plans to support SMIL and Blink is moving towards removing support as well.

Two projects at Google switched from SMIL to CSS animations and an issue we hit was that useful graphics attributes such as 'd' are not yet presentation attributes. Only attributes promoted to presentation attributes can be animated with CSS. Would it be possible to have even more attributes become presentation attributes in SVG2? Making 'd' and 'transform' presentation attributes has been discussed before but seems to have petered out. For folks not familiar with CSS animations, I created a small demo showing how they work:

The CSS Transforms spec is now responsible for defining transform. `transform` is indeed a presentation attribute today. And partly implemented that way in WebKit and Blink already.

For the `d` attribute there was a discussion about the property name. (d is not very descriptive.) There were also concerns about the semantic of such a property. Should the value be a string? Should the value be path functions as we have them for clip-path? E.g path("M100,100...."), polygon(), inset(), circle()? And I think there were even concerns about animating the string.

Anyway, d as CSS property is not off of the table. Just not a priority for the SVG 2 specification.


A second set of issues is the feature gap between web animations / css animations and SMIL. These are probably best addressed in www-syle but they are worth pointing out; Brian outlined a number of these in

Received on Tuesday, 2 June 2015 19:44:51 UTC