How long should the subscriber wait ?

Question

I am new to JMS pub/sub. I have read a number of samples. Most of them shows the subscriber uses the onMessage method to wait for publications on the topic that they subscribed to. My understanding is this will wait for ever. My questions is if the subscriber does not specify any expiry for its subscription, it will be active forever. Is there a way for the subscriber to know that the publisher will never publish the topic again or it has been deregistered by the broker due to whatever reason so that it can react to the condition instead of wating forever ? If there is a best practice, will the practice be different between a durable subscriber and a non durable subscriber ?

Answer 1

The durable and non-durable cases are kinda similar - in both a JVM starts up and a subscriber is created for the lifetime of the JVM unless the code decides to terminate the subscription.
Typically all JMS clients work like that; a consumer is started and usually used until the JVM closes or some event occurs so that the client closest the subscription.
There's not usually a big cost with keeping open a subscription. One of the wins of keeping a subscription open is that it can handle the producer coming and going; things remain loosely coupled.
If you really want to know when a producer starts/stops most JMS providers support advisory messages which you can watch to see when things start/stop.
http://incubator.apache.org/activemq/Advisory+Message
Though with advisories you need to keep a subscription open to watch for a new producer starting, so in the grand scheme of things you don't gain that much, unless you are talking about lots of different subscriptions within the same JVM.
James
http://logicblaze.com/
Open Source SOA