That is great information. The wire protocol is quite elegant. But from a users perspective the router will always see the same identity, or better, the app sending messages via router to the other n apps will always reach each of them by consistent identities.
At least this is what I'm counting on with my code. ;) -- Bruno Rodrigues Sent from my iPhone No dia 18/12/2013, às 09:20, Laurent Alebarde <[email protected]> escreveu: "*From my experience if you set the dealer identity before connecting, by design it will always use that value. *" : Except during the handcheck duration where the identity is a self determined one of 5 bytes. If you set the identity option, it will be changed by the server at the end of the handcheck, from the information it collects from the metadata. In most use cases, this is not seen by the application and you should not be aware of it, except if you connect a ZMQ_STREAM to a ZMTP socket. Whatever connexion you make, the server will use a random 5 bytes identity for the first connexion to a given socket, and will increment it for the next one. Le 17/12/2013 23:44, Bruno D. Rodrigues a écrit : >From my experience if you set the dealer identity before connecting, by design it will always use that value. It's when the identity is not set manually and the router calculates a unique id that it will change. -- Bruno Rodrigues Sent from my iPhone No dia 17/12/2013, às 22:33, Greg Ward <[email protected]> <[email protected]> escreveu: On 17 December 2013, Lindley French said: This is a bit of a novice question, but I can't find any answer to this in the Guide.... If a DEALER is connected to a ROUTER, and the underlying TCP connection goes down for some reason and is then reestablished, will the DEALER still have the same identity in the ROUTER? No. Disclaimer: I'm not a 0MQ expert, and I haven't read the source code or the documentation for this. But late last week, I incorrectly assumed that the same 0MQ ID works for the entire lifetime of a server, even if clients disconnect and reconnect. That incorrect assumption cost me 2 days of debugging. Learn from my fail. ;-) Greg _______________________________________________ zeromq-dev mailing [email protected]http://lists.zeromq.org/mailman/listinfo/zeromq-dev _______________________________________________ zeromq-dev mailing [email protected]http://lists.zeromq.org/mailman/listinfo/zeromq-dev _______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
_______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
