Ok cool that sounds good. But there's still something else going on with the tuscany schema's that i've not tracked down as there's no validation errors even without any jsonp schema at all, and also if i add <tuscany:binding.foo /> to a testcase composite the testcase still runs fine its only when using the sca namespace like <sca:binding.foo /> that the testcase fails.
...ant On Sun, Aug 9, 2009 at 6:46 PM, Raymond Feng<[email protected]> wrote: > The xsd is picked up by Tuscany and it is passed to the > SchemaFactory.newSchema() method. Now we have two XSDs, one from > assembly-xsd (tuscany-sca-1.1.xsd which includes other xsds) and the other > is binding-jsonp.xsd. > > SchemaFactory.newSchema is keeping a cache based on target namespaces. When > the binding-jsonp.xsd is loaded, there is already entries from > tuscany-sca-1.1.xsd that has the same namespace (tuscany namespace) and > binding-jsonp.xsd is ignored. > > We can fix the problem as follows: > > For all the Source passed to newSchema(), we read the targetNamespace first > and group the XSDs by tns. If we see multiple files have the same namespace, > then we generate a façade XSD that includes the original XSDs. > > Thanks, > Raymond > -------------------------------------------------- > From: "ant elder" <[email protected]> > Sent: Sunday, August 09, 2009 3:59 AM > To: <[email protected]> > Subject: Re: Adding xsd's for scheme validation of extensions > >> On Sat, Aug 8, 2009 at 5:33 PM, Raymond Feng<[email protected]> wrote: >>> >>> You can plug it in using >>> >>> META-INF/services/org.apache.tuscany.sca.contribution.processor.ValidationSchema. >>> Each text line in the file can point to an XSD using its resource name. >>> We >>> might need to tweak the following method so that xsd:import/xsd:include >>> across modules can work. I can help to fix it if you see problems. >>> >>> >>> org.apache.tuscany.sca.contribution.processor.DefaultValidatingXMLInputFactory.resolveResource(String, >>> String, String, String, String) >>> >> >> Ok i've started trying this using the JSONP binding to try it out in >> r802514. There must be something else going on as it doesn't seem to >> have made much difference, the testcase in binding-jsonp-runtime runs >> without producing any scheme validation error message both before and >> after that change when the .composite is as is or changed to have >> extra unused attributes added to the binding.jsonp element. >> >> ...ant > >
