Lists Home |
Date Index |
- From: "Simon St.Laurent" <email@example.com>
- To: firstname.lastname@example.org
- Date: Wed, 29 Nov 2000 09:44:55 -0500
At 10:02 AM 11/29/00 +0000, Henry S. Thompson wrote:
> The connection between instance and schema can be established in a
> number of ways:
> 1) An application may have a schema built in, and always use that one;
> 2) An application may allow a user to provide one or more schema
> documents, e.g. on the command line, for use in validation;
> 3) An instance may contain
> attributes, which in turn associate namespaces with the location
> of schema documents for that namespace, and an application may
> exploit this to find those documents;
> 4) An application may attempt to dereference namespace URIs and use
> schema documents found as a result.
> The XML Schema draft spec. does _not_ mandate a particular order in
> which these strategies should be followed, if more than one is
> supported, but does encourage processors to support all of (2) -- (4).
And just to make the point yet again, lest anyone think Henry's word is the
final word on this issue, option #4 remains controversial and weakly
defined. Applications which expect to be able to dereference namespace
URIs and find schemas should be prepared to be disappointed. Likewise,
developers who rely on namespace URIs to point to their schemas shouldn't
expect applications to necessarily look there.
I'd suggest that we start talking about a clearer definition of what a
namespace is and what might 'live' there, but it seems to be one of those
XML Elements of Style / XML: A Primer, 2nd Ed.
XHTML: Migrating Toward XML
http://www.simonstl.com - XML essays and books