On Fri, Jan 27, 2012 at 3:05 PM, Leonid Ganeline <[email protected]> wrote:
> "...a queue device (or client)... " it is from the 0MQ Guide. Seems > for the PPP protocol it can be just Client. So we eliminate one more > role - "queue device". All discussions about Queue and Client are > irrelevant to PPP (?) Nice catch. I've cleaned up the PPP text to remove references to "queue". > 2. Maybe addition of the Role section (or whatever name for it?) with > definition of partes: Client, Service (of Worker?) and Peer. Now > roles/participants came undefined. OK, will do. > 3. Not sure, but... I think the wire format of the Command is an > implementation detal. Just names of the Commands would be enough to > describe PPP. We must fully define the frames so that different implementations of PPP can interoperate. > 4. "...If the worker detects that the queue has disconnected, it MUST > restart a new conversation. " - there is no definition of "restart a > new conversation". Seems it is sending a Ready command. If so, it > should be "...If the worker detects that the queue has disconnected, > it MUST send Ready command again.." Great. > 5. Maybe addition of the system state as: Disconnected, Connected with > difinitions - would be good for clarification. (and Section: System > states) I'd do that but am focused on other stuff to the extent that what I write will be rubbish. So if you want this, propose me text and I'll add it. Thanks for the improvements to the spec. -Pieter _______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
