Re: Some comments on the current draft

Le 24 sept. 06 à 22:01, Laurie Voss a écrit :
> 1. Semantic abdication


> This leaves classes and roles with no advantages in creating semantic
> value over custom XML brought in on another namespace, apart from
> marginally simpler syntax, and the significant disadvantages of
> overloading classes (which are supposed to carry strictly
> presentational information) and removing the utility of @role for
> accessibility, as well as complicating machine processing of
> semantics.

The "class" attribute[1] has not been created for "presentational  
information". The semantics value of an element or an attribute is  
*defined* by a specification.  It is important to not mix the notions  
of elements, attributes and values of attributes.

"Values of attributes" can be perfectly standardized in a  
specification. A few of them have been in the past. For example, see  
"section 6.12 Link Types"[2] or section "6.13 Media descriptors"[3]


> - Stop overloading classes with semantic information

do you mean 'class' attribute? Because nothing has been suggested in  
XHTML 2.0.

XHTML 2.0 proposes property attribute for semantics and RDFa  
mechanims to extend semantics.

> - Avoid @role for semantic information; give it the role originally
> intended -- accessibility -- and clearly specify a set of values that
> @role can take, giving non-visual user agents a specification to
> develop to.
> - Recommend that any semantic extension be produced via namespaced XML
> extensions. If any of these extensions become particularly popular,
> the W3C should get involved in standardizing them too.

This is opposite to what you said earlier in your mail.
	"The role of a standards body should be to standardize
	what people are already doing, possibly nudging them
	gently towards more flexible or accurate practices."

If we look at what is happening right in the HTML community, most of  
the people are using "class" attribute to extend the semantics of  
their documents more than using "namespaces".

> XHTML 2.0 needs to be released, and soon, or the W3C risks being as
> irrelevant to the actual shape of the web as it was with HTML 3.0.

I think you could accelerate the process by
	- proposing a series of test cases for each features
	- best practices for each elements, classes and attributes
	- reviewing deeply the specification and sending your comments to

As you are an employee of Yahoo!, you could ask your AC Rep, Laurence  
Mann, to join the HTML WG. I think it would be much appreciated.

See my comments for example 

Best Regards.

Karl Dubost -
W3C Conformance Manager, QA Activity Lead
   QA Weblog -
      *** Be Strict To Be Cool ***

Received on Tuesday, 26 September 2006 06:14:24 UTC