Good catch. I'll add it to the wiki page.
On Thu, Jun 26, 2014 at 10:31 AM, Goswin von Brederlow <[email protected]> wrote: > On Mon, Jun 23, 2014 at 08:06:30PM +0200, Peter Kleiweg wrote: >> Pieter Hintjens schreef op de 23e dag van de zomermaand van het jaar 2014: >> >> > On Fri, Jun 20, 2014 at 6:44 PM, Peter Kleiweg <[email protected]> wrote: >> > >> > > A socket connect for these two addresses return no error. With >> > > version 4.0 they both return the error "invalid argument": >> > > "tcp://localhost:invalid" >> > > "tcp://in val id:1234" >> > >> > That's strange, and not normal. I'll investigate. >> > >> > > The Go version of test_security_curve fails. >> > >> > This could be due to various things. I don't think we modified the >> > curve test case since 4.0. Can you investigate? >> >> test_security_curve.cpp is identical for 4.0.4 and 4.1.0 >> (except for comments). >> >> >> test_connect_resolve.cpp is different. The two invalid >> tcp-addresses above are supposed to return an error in 4.0.4, >> but not in 4.1.0. > > The behaviour has changed when authentication (ZAP) fails. The > connecting socket then has no outgoing pipe causing zmq_msg_send() > behaviour to change (blocks forever iirc). > > In the libzmq tests a send timeout was added to the bounce helper to > handle this. > > MfG > Goswin > _______________________________________________ > 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
