- From: Kay, Michael <Michael.Kay@softwareag.com>
- Date: Fri, 6 Dec 2002 13:27:16 +0100
- To: Richard Tobin <richard@cogsci.ed.ac.uk>, John Cowan <jcowan@reutershealth.com>, mrys@microsoft.com
- Cc: pgrosso@arbortext.com, Michael.Kay@softwareag.com, w3c-xml-core-wg@w3.org, w3c-xml-query-wg@w3.org, xml-names-editor@w3.org
> > > 1. XML 1.1 describes a true superset of XML 1.0 > > > > The set of possible XML 1.1 documents is not quite a > superset of the > > set of XML 1.0 documents, but the differences are lexical and don't > > show through at the Infoset level. > > Am I right in thinking that Michael Rys wants it to be a > superset so that one can always serialize a 1.0 or 1.1 > infoset as 1.1? Rather than having to look at the infoset to > see what it can be serialized as? > It's not just a serialization issue. We need to define the semantics of validation when applied to a Document object in the XPath data model. We do this by converting our Document object to an Infoset, and invoking schema processing on the Infoset to create a PSVI. If the Infoset uses some of the new Name characters in an ID value, this is going to need a schema processor that validates according to 1.1 rules. That's fine if we can always invoke a schema processor that handles 1.1. If we sometimes have to invoke a 1.0 schema processor, because we've got an infoset that's valid in XML 1.0 but not in XML 1.1, then we are really introducing complexities that we could do without. Michael Kay
Received on Friday, 6 December 2002 07:27:53 UTC