RE: [css3-images] Remove all references to out-of-document elements and CSSElementMap?


[Tab Atkins Jr.:]
> 
> The element() function currently allows an out-of-document element to be
> selected and (in some cases) rendered, through a mechanism that is host-
> language dependent.  There is an informative example of this using the
> CSSElementMap API I proposed to be added to HTML for this purpose.
> 
> There's been a relatively large amount of discussion and disagreement over
> multiple aspects of this:
> 
> 1. Some people don't like having id selectors refer to out-of-document
> elements, as it's inconsistent with how the same selector would work in
> other contexts.
> 
> 2. The behavior of this for SVG fragments is underdefined currently,
> because CSS in out-of-document trees is underdefined.
> 
> 3. The existence of the CSSElementMap in HTML is contested.  Some suggest
> that it should be part of CSSOM instead, or have a different name, or have
> slightly different behavior (such as relying on the element's own @id
> attribute, rather than using the key in the map as an alternative id.
> 
> 4. Some people think the current ability is too limited to really be
> useful, and we should wait and combine it with the ability to reference
> elements from external documents as well.  This would allow things like
> making a paint-server gallery in an SVG file and then referencing it in
> your CSS; right now you'd have to insert the SVG into every page, either
> as direct markup or with script.
> 
> My proposal is to drop all references to this ability for now, and resolve
> these issues either in Images 4 or another spec like CSSOM.
> I'd remove all references to out-of-document elements from the spec.
> This still leaves in useful abilities, like the ability to use a non-
> rendered element; it would just require putting the element into the
> document and giving it display:none.
> 
> This also lets us avoid the handful of issues around this subject recorded
> in <http://wiki.csswg.org/spec/css3-images>.
> 
> Thoughts?
> 
I support this.

Received on Tuesday, 20 March 2012 15:54:08 UTC