- From: Daniel Holbert <dholbert@mozilla.com>
- Date: Thu, 21 Jan 2010 12:53:58 -0800
- To: www-svg@w3.org
- CC: Robert O'Callahan <roc@ocallahan.org>, Brian Birtles <birtles@gmail.com>, Jonathan Watt <jwatt@jwatt.org>
Hi www-svg,
To follow up on my previous message, I posted a grid of testcases for
testing how various SVG implementations handle attributeType="XML" vs
"CSS" for presentational attributes:
http://people.mozilla.org/~dholbert/tests/smil/compat_tests/xmlVsCssChart_v1.svg
(There's also a ".html" version for Adobe SVG Viewer -- it wouldn't load
the plain SVG version on my machine, probably because I need to tweak
the mimetype or something.)
Here are the results from the above-linked grid, across various SVG
implementations:
| (a) | (b) | (c) | (d) |(e) | (f) | (g) | (h)
ASV | Gray | 38px | Blue | Green | Blue | Green | Blue | Green
Opera | Yellow | 18px | Green | Green | Red | Green | Blue | Green
WebKit | Red | 38px | Blue | Green | Red | Green | Blue | Green
(I'm not including Mozilla/Firefox because it doesn't support animating
presentational attributes yet -- that's what I'm working on & why I'm
writing this message)
VERSION INFO:
Opera 10.10 on Ubuntu 9.10
Safari 4.04 on OS X
Adobe SVG Viewer 3.0 on IE 6 (via Wine, on Ubuntu 9.10)
DISCUSSION:
ASV matches "Behavior 1" from my previous message -- it follows the last
paragraph of SVG 1.1 section 6.4 exactly. (completely ignoring the value
of "attributeType" for animations on presentational attributes.)
Opera matches "Behavior 2" from my previous message -- it treats the XML
attribute and the CSS properties as completely distinct entities. (with
the CSS property's value overriding the XML attribute, whenever both are
set or animated.)
WebKit is a sort of strange intermediate.
*Aspects that match Behavior 1: When the property & attribute are
simultaneously animated, they interact addititvely with each other (see
(b)). Also, the property doesn't automatically "win" when there's a
conflict (see (c)).
*Aspects that match Behavior 2: Animations on the XML attribute *do*
"lose" when there's an *inline style* set (see (e)).
So, there's apparently no clear agreement among SVG implementations
about what "the right thing to do" is here...
Received on Thursday, 21 January 2010 20:54:31 UTC