On 01/17/2011 04:28 PM, Dennis Sosnoski wrote:
> ...
> I'm now wondering, though, if I'm better off just bypassing the whole
> Server/Control logic in these tests and instead just starting an
> in-process service instance directly in my test code. That way I can
> control the bus configuration and don't need to play any games with
> passing data. Anyone see any drawbacks to this approach for server-side
> testing?
>   

I gave this a try, but I've got a problem during the RM sequence
creation. The request goes out and the successful response comes back
(as shown by console logging), but the
org.apache.cxf.ws.rm.Proxy.createSequence() call returns null. There's
an interesting comment in this method, BTW: "// tried using separate
thread - did not help either". It's nice to see an occasional comment in
the code, but that one is a bit scarey.

I traced this further to the response message handling, where
Proxy$RMClient.processResult() is being called and expecting to get the
message data as a List. The MessageImpl has a null for the
java.util.List entry in defaultContents[], so it returns null.

Anyone have an idea why this might be messing up, given that both the
request and the response message look ok?

Thanks again,

  - Dennis

Reply via email to