Extensional interpretation of rdfs:subClassOf?

In message [1], Peter F. Patel-Schneider says:
[[
As far as the logic underlying OWL is concerned they are exactly the same.

   :A rdfs:subClassOf :B
   :B rdfs:subClassOf :A
entails
   :A owl:equivalentClass :B

and

   :A owl:equivalentClass :B
entails
   :A rdfs:subClassOf :B
   :B rdfs:subClassOf :A
]]

I understood that Owl classes were defined extensionally, and would 
therefore expect that owl:equivalentClass was an extensional equivalence, 
and that RDF classes are defined intensionally.

Does Owl strengthen the interpretation of rdfs:subClassOf?  It seems that 
would be legitimate under the semantic extension regime set out in the 
semantics specification (True in RDF remains True in Owl), but I wonder 
about the wisdom (and potential for interoperability failure) if the 
interpretation of RDF core vocabulary can be modified by other specifications.

#g
--

[1] http://lists.w3.org/Archives/Public/www-rdf-logic/2004Jan/0030.html


------------
Graham Klyne
For email:
http://www.ninebynine.org/#Contact