Re: A release for JavaOne?

[Jeremy Boynes]

Raymond Feng wrote:

Do we want to differentitate the Threading-based async and 
Messaging-based async?


From a component programming model I don't think there's a need to 
differentiate. I mean, the component should not be aware of what 
transport is being used. The component would specify the quality of 
service it required and the runtime would be free to choose any 
implementation that satisfied those requirements.

I do think there will be different (transport) binding implementations 
for the two and that the runtime will need to be able to differentiate.

I get nervous when folk mention "threading-based" approaches. I think 
both are message-based but with different qualities of service. In both, 
processing a message is work that should be scheduled by a work manager.
Specifically, neither the JMS implementation or the in-memory async 
implementation should be creating their own threads for this.

--
Jeremy