Hi Pieter, > For inproc connect works synchronously and requires that bind has > happened already. See http://api.zeromq.org/3-2:zmq-inproc#toc4 > > It's not consistent with tcp:// and ipc:// but it does work in most cases.
The odd thing I'm seeing though is that If my first connect() happens to beat the first bind(), - failing as expected - it appears that all the future attempts to connect() fail too, even though most of them certainly are happening after the bind() has had plenty of time to complete. If I ensure bind() happens first by using a small sleep() after spawning the thread which does the bind(), then it all works fine as you say (and I'm running with this workaround for now). So it's just this question of whether I should be able to retry an initially failing connect() until it succeeds (assuming the bind() completes). My test program recreates the scenario with a retry-loop, but I can't seem to get the connect() to work at all if it failed once, but I'm quite new to zmq and might be making some basic error. Thanks, Andy > > -Pieter > _______________________________________________ > zeromq-dev mailing list > [email protected] > http://lists.zeromq.org/mailman/listinfo/zeromq-dev -- Andy Ballingall Senior Software Engineer The Foundry 6th Floor, The Communications Building, 48, Leicester Square, London, WC2H 7LT, UK Tel: +44 (0)20 7968 6828 - Fax: +44 (0)20 7930 8906 Web: http://www.thefoundry.co.uk/ The Foundry Visionmongers Ltd. Registered in England and Wales No: 4642027 _______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
