ISSUE-2082 (attributeType auto): attributeType auto [Last Call: SVG 1.2 Tiny ]

ISSUE-2082 (attributeType auto): attributeType auto [Last Call: SVG 1.2 Tiny ]

http://www.w3.org/Graphics/SVG/WG/track/issues/2082

Raised by: Doug Schepers
On product: Last Call: SVG 1.2 Tiny 

Dr. Olaf Hoffmann
<http://lists.w3.org/Archives/Public/www-svg/2008Oct/0002.html>:
[[
in 16.2.5 attributeType auto is defined:

##
"auto"
The implementation should match the 'attributeName' to an attribute for the
target element. 
The implementation must first search through its list of supported CSS
properties for a matching property name (all properties supported by the
implementation, not just those defined by SVG), and if none is found, search
the default XML namespace for the element.
##

1. Surprising (already in SVG 1.1) in comparison with the types XML and CSS
is, that it is not explictly required, that 'The attribute must be defined as
animatable in this specification'.
Is this intended? My guess is not, especially because for any attribute or
property is anyway already mentioned, whether it is animatable or not. 
However that this is mentioned for XML and CSS, but not for auto is
inconsistent and a little bit confusing for the reader.

2. If for example a viewer like Opera supports CSS:width and CSS:height, 
the definition seems to suggest, that in case of attributeType auto there is
no visible animation effect for SVG:width and SVG:height, because as far as I
understand CSS:width and CSS:height have no effect on the SVG elements,
SVG:width and SVG:height are applicable for.
My impression is, that this is neither useful nor intended.

Taking into account 1. and 2. especially '(all properties supported by the
implementation, not just those defined by SVG)' results in a situation, 
where it is not predictable for an author anymore, whether an animation 
will have a visible/intended effect in (later) implementations or not, because
the author cannot completely predict, which properties outside of SVG any
viewer may support, which may collide with the name of an SVG attribute,
therefore attributeType auto or no specified attributeType becomes useless 
or even unpredictable for authors.
Assuming that this is not intended, this should be avoided/clarified/fixed in 
the draft.
]]

Received on Wednesday, 1 October 2008 14:03:17 UTC