- From: Kasimier Buchcik <K.Buchcik@4commerce.de>
- Date: Wed, 05 Oct 2005 21:41:52 +0200
- To: Mark Birbeck <mark.birbeck@x-port.net>
- Cc: ML-public-schemata-users <public-schemata-users@w3.org>
On Wed, 2005-10-05 at 20:00 +0100, Mark Birbeck wrote: > Kasimier, > > > The part about chameleon schemata seems to be incorrect on that page. > > Chameleon schemata are schemata with no @targetNamespace, > > which are <include>d into a schema with a @targetNamespace. > > The chameleon mechanism is not intended for <import>ed schemata. > > My document is actually quite out of date (although I'm not sure I agree > with *exactly* how you have worded that). > > When I was first doing this I was testing only with MSXML, which actually > allows what I have described (an xs:import with a @namespace value, to > import a schema with no @targetNamespace). However, when I later started > testing with other validators I discovered the error of my ways ;) and have Yes, the @namespace of an <import> needs to match the @targetNamespace of the <import>ed schema document, or, if @namespace is ommitted, then the <import>ed schema document must not have a @targetNamespace. > since taken a different approach (I have created 'proxy' schemas that allow > a chameleon schema to be imported into a host language using a different Yes, that's the way to go. However, using the term "imported" can still be misleading to others, since chameleon schemata are <include>d not <import>ed. > namespace to that of the host language). > > I've been working with Shane on getting some of these techniques into the > XML Schema Modularisation draft, so I haven't yet got round to updating the > Wiki--so apologies for that. np By the way, with chameleon schemata, you are hitting a wobbly spot of the XML Schema 1.0 spec. In this area, some schema processors work differently than others do. Regards, Kasimier
Received on Wednesday, 5 October 2005 19:48:43 UTC