[Role Module] Namespace for 'role' Attribute

Hi, XHTML2 WG-

I believe I've mentioned this in emails to this list before, but I 
wanted to make sure you had a Last Call comment about it (my apologies 
for it being a bit late, I didn't realize the deadline was coming up so 
soon).

The  XHTML Role Attribute Module [1] states in section 2.2. ("Host 
Language Conformance") that, "the attribute MUST be referenced using its 
namespace-qualified form (e.g., <myml:myelement 
xhtml:role='definition'>a term</myml:myelement>)." [2]

An earlier version of the spec [3] did not have this constraint, stating 
instead that, "If the host language does not incorporate the XHTML Role 
Attribute Module attribute into its own namespace, the document MUST 
contain an xmlns declaration for the XHTML Role Attribute Module 
namespace".  While this is a conformance statement on the document, it 
seems to have implications for host language conformance as well, 
implying a more relaxed approach that allows a host language to 
incorporate the module directly, without need for the namespace 
declaration and prefix.

I have no problem with requiring namespaces per se, but I don't see the 
rationale in this instance.

In the case of complex specs with intertwingled aspects and multiple 
internal dependencies, or a large number of elements or attributes, and 
a commensurately complex or large schema/DTD, there is clear 
justification for requiring that the whole be used, and maintained in 
its own namespace.

But the 'role' attribute is only one attribute and its value type, and 
even that type is ultimately meant to be spun off into its own spec, 
CURIEs.  I don't see the pragmatic or even theoretical gains for 
referencing the XHTML NS for a single attribute (in SVG, for instance). 
  If I'm missing something, please let me know what it is.

I believe that the use of the XHTML NS should be optional in the case of 
host language conformance.  There are good reasons for giving the option 
of a separate NS, so that it can still be used by authors in languages 
that don't support it directly; but if a language wants to support it 
directly, I think there should not be the namespace requirement, merely 
the stricture that the host language should not change or add to the 
semantics or conformance criteria of the 'role' attribute, and a 
normative reference to the XHTML Role Attribute Module.

(Disclaimer: Please note that this is my own opinion, and doesn't 
necessarily reflect that of the WGs that I'm in, or the W3C in general.)

[1] http://www.w3.org/TR/xhtml-role/
[2] http://www.w3.org/TR/xhtml-role/#hostconf
[3] http://www.w3.org/TR/2006/WD-xhtml-role-20060725/#docconf

Regards-
-Doug Schepers
W3C Staff Contact, SVG, CDF, and WebAPI

Received on Tuesday, 30 October 2007 04:40:35 UTC