Re: SVG and MathML in text/html

On Mar 10, 2008, at 04:12, Dan Connolly wrote:

> This is closely related to what I had in mind for
> ACTION-11 Show how <div aria="something"> works with URI based  
> extensibility

I think the SVG/MathML issue is very different from ARIA. ARIA is now  
designed in such a way that it doesn't require parser-level changes in  
order to be implementable. That is, the implementation doesn't need  
URI-based indirection. Indeed, ARIA is well on track for  
implementation in Firefox 3, Opera 9.5 and IE8. Having three of the  
top four browsers on the same page is remarkably good.

Enabling SVG/MathML in text/html, on the other hand, does require  
parser-changes and the existing above-DOM implementation expects the  
elements to be in namespaces.

> I might have missed the boat in the aria context; maybe
> I/we should open a separate issue for namespaces...
> or SVG/MathML... I'm not quite clear where to draw the
> line. Any suggestions?

I suggest we don't consider SVG or MathML as extensibility issues but  
as parts of the Web platform proper. (Likewise for ARIA.) Furthermore,  
I suggest not lumping SVG/MathML and ARIA together, since ARIA is now  
designed to work without namespaces or parser-level changes and is  
about to ship in Firefox 3, Opera 9.5 and IE8 but SVG/MathML already  
have a namespace legacy and will necessarily need parser changes that  
realistically won't make it to the upcoming round of major browser  
releases (except for the new ActiveX hooks in IE8).

So concretely, I suggest that we
  * consider ARIA a separate issue from SVG/MathML integration and  
syntax-wise done (and about to ship)
  * focus on integrating SVG/MathML into text/html as non-extension  
parts of the open Web platform
  * avoid getting carried away with general extensibility at the  
expense of the specific languages that we know in advance (SVG and  
  * consider the DOM-level node namespacing of SVG and MathML a legacy  
  * not consider the xmlns syntax as a necessary mechanism for putting  
the nodes in those namespaces

Henri Sivonen

Received on Monday, 10 March 2008 08:54:43 UTC