Sanjiva, this is when the wsdl is missing....and the code was generated using w2j/xmlbeans.
-- dims On 1/9/06, Sanjiva Weerawarana <[EMAIL PROTECTED]> wrote: > On Mon, 2006-01-09 at 00:27 -0500, Davanum Srinivas wrote: > > Sanjiva, > > > > If we generate xmlbeans databinding code, we still end up calling JAM > > to compose a schema. Which is wrong! > > +1. > > > IMHO. We should get the xsd from > > the xmlbeans generated code itself. > > Why not get it from the WSDL? In the scenario you're describing, there's > a skeleton and a WSDL, right? In that case, why not get the XSDs from > the WSDL itself? > > > See Axis 1.X's XmlBeanSerializer's > > writeSchema method, it picks up the schema at runtime for a specific > > class. FYI, the WSDD reference has it under beanMapping/typeMapping > > for each specific class (not for schema as a whole) > > That's simply an association from Java <-> XSD. I'm fine with storing > that if we need it, but we were talking about XSD *namespaces* and not a > mapping. > > > As for ADB, there is a namespace for a specific bean, but there is no > > way to get that information at runtime. So either we need a mechanism > > to "get" the schema information from the ADB generated classes or we > > have to keep it in services.xml. > > Again, the WSDL has it right? > > The problem seems to be that we're *losing* the WSDL when we codegen a > skeleton. Instead of losing it and trying to replicate that data in > services.xml, let's copy the WSDL at skeleton generation time into the > right place (as service.wsdl or whatever the right name). That way we > don't lose information we already had and then try to recover it by > adding more gorp to services.xml. > > Sanjiva. > > > -- Davanum Srinivas : http://wso2.com/blogs/
