I'm unsure about that. It is defined in the JAX-WS specification that
endpoint + "?wsdl" gives you the real-time contract that the endpoint
supports (which is a solid benefit of SOAP services), so I can't see
that reliance on that when generating artifacts is really all that
suboptimal. Indeed, generating JAX-WS artifacts using repository-kept
WSDLs that potentially diverge from what the endpoint is advertising via
?wsdl seems to carry its own risks, as you end up having two sources of
record over what the contract is. You can't just ignore the endpoint
WSDL, because AFAIK the ?wsdl address, per the JAX-WS specification, is
the authoritative contract supported by the endpoint--otherwise, it
would seem to be a CXF bug if the endpoint were publishing WSDLs that
weren't what the endpoint is actually supporting.
Glen
On 06/15/2011 05:30 PM, Christian Schneider wrote:
Absolutely when testing or learning services it is good to use the
easiest way. When building a bigger SOA though it is important to have a
clearly defined source for the wsdls.
The maven repo is the easiest way to start with a repository. It has
limitations though. For example it is currently not possible to split
wsdls in abstract, concrete and xsd parts like many like to do.
Christian
Am 15.06.2011 22:23, schrieb Gunnar Morling:
I like the idea of using Maven repos for storing service artifacts.
Nevertheless using published contracts ca be practical IMO for ad-hoc
tests and similar.
--Gunnar
--
Glen Mazza
Software Engineer, Talend (http://www.talend.com)
blog: http://www.jroller.com/gmazza
