-
30. Re: Remote replyto functionality - Bridge configuration in p
thomasra Jun 13, 2007 4:30 AM (in response to thomasra)I came to the same conclusion after looking at the activemq code.
The pun was intended, Tim. ;)
I'll see if there are any bugs registered in activemq for this issue...
The original suggestion still stands, though, regarding remote reply functionality. -
31. Re: Remote replyto functionality - Bridge configuration in p
timfox Jun 13, 2007 4:44 AM (in response to thomasra)"thomasra" wrote:
The pun was intended, Tim. ;)
NP, don't worry I am always grumpy when I first wake ;)
I have added a task
http://jira.jboss.com/jira/browse/JBMESSAGING-988 -
32. Re: Remote replyto functionality - Bridge configuration in p
timfox Jun 13, 2007 4:49 AM (in response to thomasra)"timfox" wrote:
[
I have added a task
http://jira.jboss.com/jira/browse/JBMESSAGING-988
Thomas, if you want, you could complete this task and contribute it? (hint, hint) ;) -
33. Re: Remote replyto functionality - Bridge configuration in p
thomasra Jun 13, 2007 4:51 AM (in response to thomasra)...and yes, removing the destination check allows the message to be sent.
-
34. Re: Remote replyto functionality - Bridge configuration in p
thomasra Jun 13, 2007 5:18 AM (in response to thomasra)I commented the JIRA issue.
-
35. Re: Remote replyto functionality - Bridge configuration in p
sergeypk Jun 27, 2007 7:58 AM (in response to thomasra)Actually, the JMS spec says JMSDestination is supposed to be ignored when a message is sent, the producer's destination must be used instead, and the message's JMSDestination must be overwritten with the real destination afterwards. So I think that even checking in ProducerAspect shouldn't be done.
The spec is a bit unclear since it also says that InvalidDestinationException "must be thrown when a destination is not understood by the provider". I'd take that to mean "the destination passed to createProducer/createConsumer".
Also, the code currently requires JMSReplyTo to be a JBossDestination as well, I don't know if this restriction is necessary. -
36. Re: Remote replyto functionality - Bridge configuration in p
thomasra Jun 27, 2007 8:18 AM (in response to thomasra)"sergeypk" wrote:
Actually, the JMS spec says JMSDestination is supposed to be ignored when a message is sent, the producer's destination must be used instead, and the message's JMSDestination must be overwritten with the real destination afterwards. So I think that even checking in ProducerAspect shouldn't be done.
I agree with this - this is the responsibility of the "other side" in order to be compliant - they should overwrite it with the correct destination from the producer.
However, the JMSReplyTo bit is more difficult. Like JMSDestination, it doesn't apply across providers, but AFAIK there is no API available to set this on producer, which means it cannot be JBM-specific if we want to support this. Though the most likely case is using Queue/TopicRequestor, which pretty much hides all this for the client anyway. I think temp destinations are the only way to get remote replyto to work correctly. -
37. Re: Remote replyto functionality - Bridge configuration in p
timfox Jun 27, 2007 8:45 AM (in response to thomasra)Just to clarify things here:
The change Sergey is making is not because we are not compliant.
In an ideal world ActiveMQ would fix their stuff, but I am not holding my breath.
The reason for this task is to allow us to work with ActiveMQ without having to wait for them to fix their code.
Hope that is clearer. -
38. Re: Remote replyto functionality - Bridge configuration in p
sergeypk Jun 27, 2007 8:55 AM (in response to thomasra)ActiveMQ has already fixed it, according to http://issues.apache.org/activemq/browse/AMQ-1232
I think I was misunderstood - I think we shouldn't do any checks on the value passed to setJMSDestination - neither in setJMSDestination, nor in ProducerAspect. It's simply not necessary (as far as I can tell), the spec even says we must ignore the value completely.
My question was whether we should also skip all checks on JMSReplyTo - e.g. to allow sending a JBossMessage with JMSReplyTo set to an ActiveMQ destination. -
39. Re: Remote replyto functionality - Bridge configuration in p
timfox Jun 27, 2007 9:02 AM (in response to thomasra)Yes, the point of this task is to remove those checks.
-
40. Re: Remote replyto functionality - Bridge configuration in p
thomasra Jun 27, 2007 9:07 AM (in response to thomasra)Yes, like I said: it is the "other" side that is responsible for overwriting with a correct JMSDestination. BUT, in order to support something like the topic says (remote replyto) we would have to do something about the JMSReplyTo value (JBM-specific or not).
-
42. Re: Remote replyto functionality - Bridge configuration in p
mnop629 Dec 7, 2009 11:21 PM (in response to thomasra)For the Bridging from JBM to ACtiveMQ, I saw there is a bounded property "outbound" in ds.xml. Is there any other way to bound the physical queue destination name like below:
queue/qTempBridge
It wouldn't able to resolve in JBM, it said naming not found XXXX.. -
43. Re: Remote replyto functionality - Bridge configuration in p
mnop629 Dec 7, 2009 11:23 PM (in response to thomasra)"mnop629" wrote:
For the Bridging from JBM to ACtiveMQ, I saw there is a bounded property "outbound" in ds.xml. Is there any other way to bound the physical queue destination name like below:
"<attribute name="TargetDestinationLookup">queue/qTempBridge"</attribute>"
It wouldn't able to resolve in JBM, it said naming not found XXXX.. -
44. Re: Remote replyto functionality - Bridge configuration in p
mnop629 Dec 7, 2009 11:24 PM (in response to thomasra)For the Bridging from JBM to ACtiveMQ, I saw there is a bounded property "outbound" in ds.xml. Is there any other way to bound the physical queue destination name like below:
queue/qTempBridge
It wouldn't able to resolve in JBM, it said naming not found XXXX..