I was having these questions related to how flexible
are the libraries developed in xml area.
1. Schema validation.
There are already so many schema validators, code generators from schema. Even xerces-2 has schema validator.
Now here are modules which should get integrated properly.
a. java code generated from schema should be able to hook in with xerces-2 parser so that validation and transforming to java object can happen simultaneously.
b. Integrating the schema validator to work with native XML databases (like Xindice).
2. Java to schema generators.
Again here also, there are many schema to java code generators. In order to generate these classes, it needs to parse schema document and convert it into java objects representing schema objects. Since schema itself has schema (schema's schema), are these generators (like castor) are "complete" enough to take schema's schema and generate the code representing the same classes which they used to generate it? This is like checking how "complete" an implementation is. For eg., we can use C language to write a C compiler, java to write a java compiler. Similarly we should be able to use schema generated classes to generate themselves using schema's schema.
3. Marshalling/Unmarshalling of objects.
JAX-RPC has interfaces defined for serializing and deserializing. And we have got one more spec JAXB for converting XML to java (& vice versa). Do we need 2 separate specs? JAX-RPC should be using the JAXB framework to marshall/unmarshall the java objects (passed as params).
I would be keen to know about any response...