Adrian/Christian, I've been looking over the JMSConduit code. I believe we could solve this issue if we allowed for developers to inject their own JMS listener into the conduit.
We have set up an interceptor to inject our own JMS correlation Id into the Message Object. If we had a way to inject our own consumer to the conduit we could manage the persistence and correlation of request/response messages. Let me know what you think. Thanks! -Chris Christian Schneider wrote: > > Hi Adrian, > > I think this can be done in CXF. I was talking more about what you do in > the application layer and how you design your service. > For example if your service call creates a customer object then you can > simply return a generated customer id on success. > In this case you need the correlation id to make sense of the reply. If > you instead return the complete Customer data you just set including the > customer id that was created while doing > the call then you can probably process the reply without knowing the > request that caused this reply. When doing loosely coupled asynchronous > services it is very important to design like > in the second example. > > Greetings > > Christian > > Adrian Corcoran schrieb: >> Hi Christian, >> >> So when you say that you have to write the client in a really >> asynchronous >> way are you saying you don't really think that there is a way that this >> can >> be done using CXF? >> >> Thanks, >> Adrian >> > > > -- > > Christian Schneider > --- > http://www.liquid-reality.de > > > -- View this message in context: http://www.nabble.com/jms-client-restart-tp20853766p20917773.html Sent from the cxf-user mailing list archive at Nabble.com.
