In WSDL 1.1, there are a few concepts in the model:
1) portType: defines the interface/operation
2) binding: defines the communication protocol and message format
3) service/port: defines the endpoint (address, binding) where the service
is published
binding.ws allows the configuration of @wsdlElement to point to the above
models in the WSDL using QNames. I think the JIRA complains that if the
QName cannot be resolved a model defintion in the WSDL, we should have
meaningful messages to report that. For example, if we have:
<binding.ws wsdlElement="http://ns1#wsdl.port(service1/port1)"/>
If there is no WSDL defines a port "port1" under a service "service1" with
"http://ns1"; namespace, we should report the problem instead of NPE.
Thanks,
Raymond
--------------------------------------------------
From: "Dan Becker" <[email protected]>
Sent: Tuesday, January 13, 2009 12:25 PM
To: <[email protected]>
Subject: WSDL Binding validation scenario
I'm looking into Jira TUSCANY-2754
(https://issues.apache.org/jira/browse/TUSCANY-2754), and I seem to be
drawing a blank on a good user scenario to recreate the problem. (Perhaps
not enough coffee today.)
The reporter is stating that any errors in a composite binding.ws
wsdlElement value (either namespace, service, port, or binding name)
should be validated and flagged if misspelled.
I see Tuscany has a binding-ws-wsdlgen test case which will generate WSDL
from a Java interface, and I also see Tuscany has binding-ws-xml read and
write test cases to read to and from a composite XML with embedded
binding.ws elements. However neither of these demonstrate validating a
composite wited to a known WSDL service or reference. Can someone point me
to one of the Tuscany samples or test cases that show validating the
composite binding.ws elements?
--
Thanks, Dan Becker
