Currently messages sent over the -stomp and -hybrid transports can remain in the message queues on the server for an infinite period of time.
As the user can configure a timeout to wait for a response for the message, this can clearly mean that a service could process a message even after the client is no-longer waiting for a response.
It is therefore neccesary to allow the user to specify a time-to-live for all outbound messages, e.g. TimeToLive. The client should also be modified to wait for responses for "TimeToLive + RequestTimeout" period of time.