On 07/14/2011 04:18 PM, Pieter Hintjens wrote: > Seems in some cases (pub-sub at least but probably other sockets too), > on OS/X and perhaps other BSD systems, 0MQ dies with a SIGPIPE signal. > > It's not an assertion, the signal simply appears to kills the process. > > https://zeromq.jira.com/browse/LIBZMQ-230?focusedCommentId=13309#comment-13309 > > Presumably this affects all versions of 0MQ. There's a discussion here: > > http://stackoverflow.com/questions/108183/how-to-prevent-sigpipes-or-handle-them-properly > > Is this a known issue at all?
Yes. It is a know issue. The problem IIRC is that the solution is platform-dependent (such as setting SO_NOSIGPIPE). I would have to have access to the platforms causing problems to fix it. Still, patches are welcome. Martin _______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
