Re: [Fwd: Using XMLNS in link/@rel]

Ben Adida wrote:
> Julian Reschke wrote:
>> If recipients need to change, then you have introduced a new
>> notation/syntax/whateveryouwannacallit.
> 
> Recipients only need to change if they lent @rel a
> syntax/meaning/whateveryouwannacallit beyond the spec. If you assumed
> @rel in HTML or XHTML was a URI, you went beyond the spec.

Again, the assumption, if any, is that the @rel value alone identifies 
the relation. Whether it uses URI syntax or not is totally irrelevant.

> RDFa specifically ensured that existing *specified* interpretations of
> @rel were preserved, i.e. link types. And RDFa specifically ensured that
> the list of link types could be extended in the future without
> interfering with our use of @rel including prefixed values which,
> according to the spec, you should have simply ignored as local values.

Where does the spec say that? Didn't you claim that presence of a 
@profile can license interpretation of unregistered rel values?

> I think that's what Steven means when he says we added a way to
> interpret the @rel syntax without actually changing it, because there
> *no* syntax specified other than CDATA.

And that's what is incorrect. *Because* the syntax is a plain string 
it's an incompatible change to define a new syntax here. Keep in mind 
that "foo:bar" already was a syntactically legal relation value before.

Best regards, Julian