At 22:50 +0200 2003-05-15, Johannes Koch wrote:
>which could be achieved in HTML 4.01:
><div class="address">
><div class="person">Haruki Murakami</div>
><div class="street">Omote-Sando</div>
><div class="city">    Tokyo</div>
></div>
>
>>The values of sem attribute and their meaning will be defined in a 
>>external extensible document.
>
>The values of the class attribute and their meaning will be defined 
>in a external extensible document.

Except that the problem of using the class attributes redefined in a 
new way will give difficulty to people with old documents. For 
example, if I have an XHTML 1.0 doc that had already a class="city" 
but with a different semantic context.
My take on it is that the value has to be "normalized" in a Note.

>A better approach in _X_HTML (IMHO) would be the usage of elements 
>from a special namespace for an address vocabulary:
>
><p><foo:address">
><l><foo:person>Haruki Murakami</foo:person></l>
><l><foo:street>Omote-Sando</foo:street></l>
><l><foo:city>    Tokyo</foo:city></l>
></foo:address></p>
>
>This of course would not be a fragment of a valid XHTML 2.0 
>document. But XHTML is meant to be extended like this, isn't it?

	Exactly. But :) if this solution is chosen how to define a 
mechanism which maintains interoperability.

For example, if we use other languages, should the XHTML 2.0 spec 
recommend some specs as :

	For Address, use vcard
	For geographical coordinate, use GML, etc.

It has a lot of issues but it would be beneficial for interoperability.
-- 
Karl Dubost / W3C - Conformance Manager
           http://www.w3.org/QA/ 

      --- Be Strict To Be Cool! ---