[EMAIL PROTECTED] wrote: > > Hi Edwin, > > Well I learn once more--quickly reading over code, even if it's code you > (think you) know well, is a pretty good way of misleading yourself if you > don't actually run tests... You're right of course; it's the > schemaLocation hint we pass as the systemId parameter, not the schema > namespace. > > I tend to think this behaviour is unfortunate, because I do believe that > the schema namespace is very useful information to an application's entity > resolver. And clearly this is something we'll have to expose when > considering the Xerces2 grammar caching design, because it's by namespace > and not by schemaLocation hint that we index grammars. But let's not open > that Pandora's box again for the moment. :-)
One aspect about namespace names (URIs) which are not equivalent to PublicIDs is that they do not necessarily map one to one with schemas (as pointed out by Kohsuke). One example is XHTML which has 3 schemas that map to the same namespace name but has 3 different publicIDs. -Edwin --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
