Ian Hickson wrote:
On Thu, 16 Jul 2009, Julian Reschke wrote:
  
A normative requirement that refers to an undefined set of specs?
    

It's just making explicit what is often implicit.

XHTML didn't say RDFa was allowed. But an RDFa spec was written that 
extended XHTML such that validators now support XHTML+RDFa.

Technically, this is a violation of XHTML.
  
This is someone off topic, but no - it is not at all a violation of XHTML.  XHTML M12N defines the XHTML Family and the characteristics one needs to be a member of that family[1].  It's not an exclusive club.  You should join!  In particular, M12N defines the mechanism whereby any spec can define additional markup and incorporate that markup into the content model in a way that is consistent with the XHTML Family view and that will work in XHTML Family conforming user agents.  This is also explicitly permitted by the relevant RFC - 3236 [2]
With HTML5, the spec explicitly allowed this.

  
[1] http://www.w3.org/TR/xhtml-modularization/conformance.html#s_conform
[2] http://www.rfc-editor.org/rfc/rfc3236.txt
-- 
Shane P. McCarron                          Phone: +1 763 786-8160 x120
Managing Director                            Fax: +1 763 786-8180
ApTest Minnesota                            Inet: shane@aptest.com