Guo, Yanchao wrote: > Hi Martin: > > Out setup is like this: > > We have a machine which is not inside our network, let us call it > server; we have a machine inside our network, let us call it client. The > client application will receive status from server, for that we have the > the server binding to a IP/PORT and publishing message, and the client > application connect to that IP/PORT and subscribing message. That works > fine. > > However, the client application should be able to send cmd to the server > application based on the received status. So for that, we need to have > the server bind to another IP/PORT and subscribing message, the client > application will connect to that IP/PORT and publishing message. That is > where we are having problem, We don't want the the server receive "old" > command messages.
Hm, not ditching commands is what you would usually expect. As far as I understand, your use case calls for implementing something like TTL, i.e. "if this command cannot be delivered in 2 seconds, please, drop it". Martin _______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
