- From: Jeremy Carroll <jjc@hpl.hp.com>
- Date: Mon, 2 Sep 2002 15:44:32 +0200
- To: w3c-rdfcore-wg@w3.org
At the telecon I mentioned that we had been conceiving of datatyping as a layer, whereas now we are conceiving of datatyping as "built-in". To try and give some examples, in S-A, S-B, TDL, Stake-in-the-ground, the syntaxes (Graph, RDF/XML, N-triple) were unchanged. The basic MT was unchanged. So a datatyped RDF/XML document could be interpreted by a non-datatype aware RDF processor. (Such a processor wouldn't understand the datatypes, but it would not make mistakes. Datatyping basically adds additional entailments, and additional guide to the applications). Now, however we have moved to a position in which the graph itself is changed and hence all other syntaxes and the MT MUST ab initio understand datatyping. There will not be any possibility of supporting RDF 2002 without datatyping. Thus the normative content of the datatyping spec is necessarily normative content of the syntax, abstract data model, and model theory specs. It is not clear to me that the non-normative content alone justifies another document. Jeremy
Received on Monday, 2 September 2002 09:46:24 UTC