- From: Bruno Chatel <bcha@chadocs.net>
- Date: Sun, 13 Jan 2002 14:09:32 +0100
- To: "Henry S. Thompson" <ht@cogsci.ed.ac.uk>
- Cc: <xmlschema-dev@w3.org>
> > I have the following schema : > > > > s1.xsd : targetNameSpace ns1 > > importing s2.xsd with NameSpace ns2 > > importing s3.xsd with NameSpace ns1 (the "main" targetNameSpace). > > > > Is this modularization strategy allowed ? > > Yes. It was my opinion too. > > It seems that Xerces does not support this kind of schema. > > In what sense, not support? If it gives an error for the nested > import, that's a bug. If it declines to take the s3.xsd hint, that's > not a bug. XSV wouldn't either. s3.xsd is not read by the TraverseSchema class when processing import element. (note : the version of xerces used is xerces_j 1.4.3). It seems to be a bug. > > > Indeed, in the GrammarResolverImpl class, a schema is referenced by > > its NameSpace (and not its location), then s3.xsd is not imported > > because xerces considered that ns1 is previously processed (by > > s1.xsd). > > That's allowed by the REC: schemaLocation on import is a hint. > > Why not include s3.xsd from s1.xsd? Because s2.xsd is used "single" (directly for some documents). Thanks ! -- bruno -- Bruno Chatel Tel : (+33)[0] 4 96 11 14 57 Email : bcha@chadocs.com ----------------------------------------
Received on Sunday, 13 January 2002 07:56:15 UTC