Unfortunately the C++ broker implementation slightly differs as the key is hard coded to "qpid.LVQ_key".
Other than that it should work.
I think the c++ side should behave the same as the Java broker and allow to specify the key.
Another reason why I didn't want to use the existing test is bcos I wanted to write a test using the new addressing syntax.
So I just wrote a single test method, it would have been nice had I been able to reuse the existing test.
Also the fact that the ConflationQueueTest is placed under o.a.q.server.queue gives the impression that it's meant for the java broker.
Once the C++ broker allows to specify the key, then we should use a single test, but use the new addressing syntax.
As a side note, we should also ensure that the systests only contains test cases that tests a broker from the client side allowing us to use them against any broker (barring version issues).
I believe all the tests in systests are like that, all though we have name spaces like "sever" ..etc.
I am not too worried about the package names, just want to ensure that we use systests for what it's intended for.