Ah, thanks for clearing that up. And it looks like there's no way to manually close/reopen the socket that gets created by the flowgraph correct?
Cameron On Mon, Aug 10, 2020 at 6:15 PM Jeff Long <willco...@gmail.com> wrote: > Also, time is always set to 0 for the setsockopt ZMQ_LINGER call, which > would cause immediate shutdown of the socket. This is not related to > timeout, which is used as the polling timeout. > > On Mon, Aug 10, 2020 at 7:06 PM Jeff Long <willco...@gmail.com> wrote: > >> At socket shutdown, LINGER determines how long close(2) or shutdown(2) >> will block waiting for queue messages to be sent. See man socket(7). >> >> On Mon, Aug 10, 2020 at 7:00 PM Cameron Matson <ncmatso...@gmail.com> >> wrote: >> >>> Hi all, >>> >>> Can someone help me understand what's going on with any of the ZMQ >>> Message Sink blocks? The block takes a timeout parameter which is assigned >>> to d_timeout, but ultimately it looks like the zmq.LINGER option (which I >>> believe is how long zmq will block before dropping the frame) is always set >>> to a different variable, time, which is initialized in the constructor to >>> be 0 that is used for the setsockopt call. >>> >>> if (major < 3) { >>> d_timeout = timeout * 1000; >>> } >>> d_context = new zmq::context_t(1); >>> d_socket = new zmq::socket_t(*d_context, ZMQ_REP); >>> int time = 0; >>> d_socket->setsockopt(ZMQ_LINGER, &time, sizeof(time)); >>> Am I missing something? >>> >>> Thanks, >>> Cameron >>> >>>