[rabbitmq-discuss] FW: Multiple consumers

Sun Jul 29 13:19:14 BST 2007

Matthias

>
> Sorry about the confusion, I meant
>   access_request(Channel*Pid*, Realm)
> So yes, the user would need to keep hold of the ChannelPid, just as they
> currently keep hold of the object reference to the Channel object in the
> Java API.
>
> This would work for both the network and direct API.

OK, if you're happy with that approach, I'll back out the named
process stuff and switch over to the user passing the pid, which I
think is generally more elegant as well.

I do have a couple of questions about this though:

1. In the direct case, the host and channel would be set in the
start() call, so just passing the Pid is fine. In the network case,
you would only start one client process, so you would still have to
pass in the channel number as an integer for the client process to
differentiate. Or would you somehow initialize a socket connection and
then pass this to a process that you spawn per channel?

2. Should the processes be linked at all? That is, if a channel
process dies, should the user process die as well?

Ben