[MOBY-dev] JAX-WS BioMoby API (2nd round).
Paul Gordon
gordonp at ucalgary.ca
Wed Jan 16 14:53:45 UTC 2008
> So how you handle famous "soap-encoding"?
> I mean that Axis 1.x is the only API which handle Moby on a server
> side so far.
> As I commented in various occasions Moby clients ignore WSDL file and
> name parameters differently.
> Another JAX-RPC API (SUN, JBoss) failed with Moby requests.
I create the "data" tag manually in the SOAPMesssage. Only takes a
couple of lines, the rest is standard use of Service and Dispatch.
>> Version 5 and version 6 don't like to live together in jMOBY
> Actually JAX-WS can be installed apart to use with JDK5.
> JEE5 specification uses JDK5, but makes JAX-WS obligatory. If fact
> this is what I'm doing using JBoss 4.2.2 (JDK5 + JBoss-WS/JAXB).
I know, but I can't create the data tag using a QName with the version
of JAX-WS on the Sun Web site. The Java SE6 version allows me to (this
is needed to declare the xsd namespace, since createName does not
stupidly allow manual declarations in the xmlns namespace). It's funny
how javax.xml.ws doesn't work the same in both versions...if you can
figure out how to declare a namespace manually in a SOAPMessage without
QName, let me know!
>> I'd also suggest that even with your own marshalling/unmarshalling,
>> you could easily reuse the existing API classes for data instances
>> that users are familiar with...
> I think I didn't explain in well. There is no "my own
> marshalling/unmarshalling". The generated classes are the JAXB
> annotated Java classes.
Okay, but when you work without the data types, JAXB creates an
AnyMobyObject? Or how does that get created? Could it use the
org.biomoby.shared.data classes instead? Sorry, I'm not familiar with
how that datatype-less way works...
> <snip/>
> My vision is that all Moby Datatypes must have a schema representation.
As far as I can recall, this is not possible (others can correct me if
I'm wrong, or these have been solved) . The main sticking points were
that if an object is a Moby Primitive in a has relationship, it cannot
be represented in XML schema. Also, I think that the fact that we must
be able to accept subclasses of declared classes, even if we don't know
about them, causes another problem.
> My generated classes being completely Moby compatible are in fact JAXB
> classes, so one can easy obtain the schema through schemagen.
<snip/>
> The JAXB generated classes has the same hierarchy structure as MoSeS
> ones (both generated from the same ontology).
I don't actually use MoSeS, so I don't have much to add here. The
generator code is Martin's domain of expertise, so I'll let him comment
on that if he likes. :-)
Regards,
Paul
More information about the MOBY-dev
mailing list