- From: Julian Reschke <julian.reschke@gmx.de>
- Date: Fri, 27 Feb 2009 23:19:14 +0100
- To: Ben Adida <ben@adida.net>
- CC: noah_mendelsohn@us.ibm.com, Henri Sivonen <hsivonen@iki.fi>, Mark Nottingham <mnot@mnot.net>, HTMLWG WG <public-html@w3.org>, RDFa mailing list <public-rdf-in-xhtml-tf@w3.org>, public-xhtml2@w3.org, "www-tag@w3.org WG" <www-tag@w3.org>
Ben Adida wrote: > Julian Reschke wrote: >> Acknowledge that there is a problem (CURIE as rel value), and resolve >> it, instead of pretending it's not. > > If there is a problem, I disagree that it is with CURIEs in @rel in > XHTML1.1+RDFa. I believe the problem is in trying to interpret a raw > link-type string without knowing where it came from. > > HTML supports @profile, which modifies the meaning of @rel values. > @profile="http://bens-crazy-parser.com" could easily process > rel="(:dc:)title" into http://purl.org/dc/terms/title. HTML4 effectively > has a default @profile, with a number of pre-specified link-types [1]. > > XHTML1.1+RDFa effectively has a default profile, too, where @rel values > are interpreted as CURIEs (into URIs, of course.) > > In other words, any assumption one is making about generically parsing > @rel without its language context is, in my opinion, strongly misguided. > @profile can and does alter that interpretation. So if @profile can > alter it for HTML4, then certainly the version of HTML has an effect on > how @rel is interpreted. You can't interpret @rel blindfolded. But I'd like to. WRT profile; I see how this can work for a single profile URI, but it does not scale, so I'm not sure how this is supposed to help with mixing relations from several profiles (namespaces) in a single document. > As I suggested in a previous email, one simple way to resolve this > potential conflict with the link-type RFC is to specify, in the > link-type RFC, that while the semantic value of a link-type should be a > URI, its syntax may be language-dependent. This shouldn't be > controversial, because this is already the case given @profile. > ... No, that's fine. What's controversial is how many different language-dependent interpretations there should be for the HTML family of languages. I think that number needs to be 1. BR, Julian
Received on Friday, 27 February 2009 22:20:09 UTC