- From: Shane McCarron <shane@aptest.com>
- Date: Tue, 26 Jun 2007 09:13:29 -0500
- To: Ben Adida <ben@adida.net>
- CC: Keith Alexander <k.j.w.alexander@gmail.com>, Cédric Mesnage <cedric.mesnage@lu.unisi.ch>, RDFa <public-rdf-in-xhtml-tf@w3.org>
Ben Adida wrote: > Okay, but if you use the RDFa DTD, then you clearly intend RDFa, right? > Otherwise, you're saying that @profile is the only way to provide > semantics. That seems a bit too GRDDL-centric a view. See below: I think > existing HTML DTDs provide plenty of semantics already. > I fear we are conflating two or possibly three things. Ben, you keep talking about HTML. In the context of this discussion I think sometimes you mean HTML4 and sometimes you mean XHTML 1.1. Let's look at both: HTML 4(.01) has a variety of inherent semantics. These are defined by its specification. HTML 4(.01) has an implementation in a DTD. The DTD conveys no semantics. It is a collection of syntax rules. A document that has a DOCTYPE that references an HTML 4.01 FPI is saying "I am an HTML document". That's great. The HTML 4.01 DTD is by its very nature extensible, and I have created an extended version of the HTML 4.01 Transitional DTD as an example. The fact that the implementation is done in a DTD has nothing whatsoever to do with the semantics. Note that there will NEVER be an XML Schema implementation of HTML 4.01 nor of HTML4.01+RDFa, since HTML 4.01 is not an XML dialect. XHTML 1.1 has an XML DTD and an XML Schema implementation. These are based upon XHTML Modularization. I have prepared M12N-compatible modules for RDFa using both XML DTD and XML Schema. I have also prepared an implementation of XHTML+RDFa using XML DTDs. Again, no semantics. Syntax. In due course, I will also provide an XML Schema implementation of XHTML+RDFa. However, again, an XML Schema implementation has NO SEMANTICS EITHER. Its just syntax rules. The semantics are defined in the prose specification. As to how we know some document contains RDFa annotation that can be transformed into RDF.... Personally I think that the DOCTYPE is an excellent indicator, assuming we provide a well known DOCTYPE name. A well known profile can't hurt, but DOCTYPEs are well understood, and browsers know to look at them, and validators do too. -- Shane P. McCarron Phone: +1 763 786-8160 x120 Managing Director Fax: +1 763 786-8180 ApTest Minnesota Inet: shane@aptest.com
Received on Tuesday, 26 June 2007 14:14:12 UTC