Details
Bug
Closed
Major
Description
---------- Forwarded message ----------
From: Michael Vorburger <michael.vorburger@ipt.ch>
Date: Jul 21, 2006 12:37 PM
Subject: Fault Classes with wsgen JAXWSProfile
To: user@xfire.codehaus.org
Hi,
How are Faults from a WSDL supposed to be correctly translated by wsgen with JAXWSProfile? I think there is a) a bug in XFire, b) an "ease of use" issue due to class naming in XFire, or JAX-WS, or my understanding:
a) If you do NOT customize the WSDL package (via package option in wsgen) and the XSD package (via <schema> <annotation> <appinfo> <jaxb:schemaBindings> <jaxb:package>) the generated code does NOT compile, the class generated from the Fault is wrong, and there is a build problem in the ObjectFactory. Found http://jira.codehaus.org/browse/XFIRE-421 about this, but 1.1.2 does NOT behave like described there. My sample project demonstrating this is at http://www.vorburger.ch/blog1/xfire112-examples-wsgen-jaxws-spring__3_default-packages_Fault_Build_Problem.zip
b) If you do customize WSDL+XSD packages, you get two Fault classes in the two different packages - but with the same name. The code compiles, but is somewhat confusing to use, leading to code like below. There appears to be no way (with XFire's wsgen/JAXWSProfile) to make those have different class names, both are named after the Element name in the XSD??
org.something.services.hello.dto.HelloFault fault =
new org.something.services.hello.dto.HelloFault();
fault.setXXX(...);
throw new org.something.services.hello.ws.HelloFault("Failure", fault);
Regards,
Michael Vorburger, IPT.ch
---------- Forwarded message ----------
From: Dan Diephouse <dan@envoisolutions.com>
Date: Jul 26, 2006 1:57 AM
Subject: Re: [xfire-user] Fault Classes with wsgen JAXWSProfile
To: user@xfire.codehaus.org
Hi Michael,
It seems we aren't following the spec quite right here. I think in this case we should name the exception class with "Exception" appended at the end. Could you file a JIRA issue for this? Thanks,
- Dan
I fixed this issue now in SVN I believe. AFAIK we follow the spec to the 't'. Granted there are many things unspecified in the spec, but in those cases we follow the RI.
Thanks for referencing this in your example, otherwise I would've missed it.