OK, so in your modified code, when you get NULL back from a recv call, you have to exit the loop. This is how the blocking recv functions in CZMQ signal an ETERM situation. zctx_interrupted will not be raised on all threads, only the one that caught the Ctrl-C.
On Mon, Sep 2, 2013 at 5:19 AM, Chris Laws <clawsi...@gmail.com> wrote: > After more investigation it appears that the problem I am seeing is > associated with the thread code sitting blocked in recv until the main > thread performs the context shutdown. When the thread begins shutting down > it goes through the termination sequence of events, one of which is to set > the linger option on sockets. I think it is this action which triggers the > "Context was terminated" to be emitted to stdout. > > I modified the simple example to send a message, from main to the thread, > upon detecting zctx_interrupted. This kicks the thread out of the blocking > recv call sooner, allowing it to exit the while loop and begin shutting > down. However, I found I still needed to insert a wait, allowing the thread > to finish up, before shutting down the context to avoid triggering > the"Context was terminated" error. > > The period of time I wait for the thread to shut down is a guess. Having a > feature like thread join might resolve what appears to be a race condition > here. > > Here is a modified version of the simple example that appears to work - it > no longer raises the "Context was terminated" error to stdout on shutdown. > > > > _______________________________________________ > zeromq-dev mailing list > zeromq-dev@lists.zeromq.org > http://lists.zeromq.org/mailman/listinfo/zeromq-dev > _______________________________________________ zeromq-dev mailing list zeromq-dev@lists.zeromq.org http://lists.zeromq.org/mailman/listinfo/zeromq-dev
