The suggestion I've seen was that "embedded" schema documents should be treated just like any other schema -- you'd have to reference them explicitly via a URI (and possibly an XPath/XPointer in that URI, to explicitly indicate where the schema starts). That URI might be a relative URI reference pointing back to the current document, but you'd still need that statement that this particular element is in fact part of the schema model for the document that contains it. Of course these schema elements would also be part of the document's content; conceptually, you'd just read them twice during validation... and that means the schema for this document would have to allow for the possibility of a contained schema fragment.
Note that schema elements which *aren't* part of the actual model could live in the same document (assuming the schema permits this); they'd just be a normal part of the document's content. I don't know whether any of this would actually be useful, never mind optimal... but it's one approach to making the concept make sense. "Come back next week; we're gonna do fractions." -- Tom Lehrer ______________________________________ Joe Kesselman / IBM Research --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
