Stefano Bagnara wrote:
> > It also seems orthogonal to a change that uses one thread
> > blocking for the event, and the rest are workers.
> probably in the JMS environment we would prefer that each
> worker (consumer) do its own work without the need of a
> single consumer that dispatch the messages (e.g: consumers
> could be installed in different machines)
You misunderstand. Yes, each of those consuming threads is fine, but for
each consumer, e.g., queue reader, we could have multiple worker threads.
However, it occurs to me, in terms of transaction behavior that we would
have to look at how that is handled in the James Strachan's code. We might
need each worker to be a consumer in order for a consumer failure to result
in the message being made available to another.
Has anyone looked? It seems to me that if a consumer crashes, we'd want to
rollback the take operation so that it is available for another consumer to
take the message.
--- Noel
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
