Re: [XHTML1.1] Error in Conformance Definition document?

On Mon, 2007-04-02 at 20:07 -0500, Shane McCarron wrote:
   
> I am sympathetic to the idea of a text/html serialization... but we got 
> slapped pretty hard for doing that in XHTML 1.0.  Basically, such a 
> "serialization" ends up being a series of conventions that, if adhered 
> to, will permit well formed, valid XHTML 1.1 content to sort of work in 
> legacy user agents.  Is that what you are looking for?

Indeed not. I mean an alternative serialization that would require some
form of actual transformation, where there are conformance requirements
(/not/ "guidelines") defined for authors and user agents for how that
serialization is created and understood. Whatever WHATWG are creating
for text/html it could hardly be described as a "series of conventions".

My basic concern here is that browser developers should be able to
backport any minor new accessibility updates to XHTML 1.1 without
committing to supporting application/xhtml+xml, for example if tabindex
is expanded to match Microsoft's and Mozilla's implementation. I don't
think such changes are sufficient to force XHTML support by themselves,
but I also don't want to see their use dependent on serving XHTML 1.1 in
a way that only "sort of works".

We don't need an HTML5 from W3C or WHATWG for this task. We need a minor
revision, an HTML 4.5. It could try and fix some of the limitations with
using ARIA with text/html:

http://www.w3.org/WAI/PF/adaptable/HTML4/embedding-20060318.html#limitations

If W3C want to drive XML markup adoption, I think there are much better
ways than with XHTML 1.1, such as:

* Better specs and education for mixing multiple markup languages (SMIL,
SVG, XHTML, MathML, etc) together effectively. This is the unique
selling point of XML-based markup from the point of view of content
creators.
* Actually registering MIME types for these languages (e.g. where is
application/mathml+xml?).
* Standardizing transformation methods for backwards compatibility
(XHTML and MathML to HTML4), for example), rather than trying to hack
support through compatibility guidelines.
* Explicitly requiring support for core languages in UAAG (access to
content is an accessibility feature).
* Making XHTML 2 rock.
* More blueskies thinking about more productive authoring tools (Dave
Raggett's working on this, which is good).

--
Benjamin Hawkes-Lewis