Re: [xml-dev] If I want to use catalogs for XSDs, must document instance

[Michael Kay <mike@saxonica.com>]

OFBADDBEF8.B9640943-ON482579E6.0021A7E0-482579E6.0029832D@LocalDomain" type="cite">
That is, although an XML application isn't doing it based on the value of xmlns: attributes (as you say: "Neither should it."), it seems to me that it *is* doing it based on namespace URIs (specifically, the first member of each pair of URI references in a schemaLocation attribute value).

Am I just splitting hairs, or am I (as I suspect) missing a deeper point here?

There can be more than one schema for the same namespace. For example, there are various schemas for the XHTML namespace describing different levels of "strictness" in what they accept.

I've always thought xsi:schemaLocation (and for that matter DOCTYPE) to be rather absurd: if I don't trust a document to be valid, why should I trust it to point me to a schema that I can use for validation? Nevertheless, there are cases where it can be very valuable. In general, however, I think anyone who wants to test a document for validity ought to be responsible for deciding what they want it to be valid against.

Have you looked at NVDL?

http://en.wikipedia.org/wiki/Namespace-based_Validation_Dispatching_Language

Michael Kay
Saxonica