IRC Log


Thursday May 6, 2010

[Time] NameMessage
[08:00] sjampoo zmq_init() does not correctly handle negative io-thread numbers, it crashes instead of returning an error. Is this a known issue or should i open on Github?
[08:01] sustrik make an issue on github
[08:01] sjampoo ok
[08:01] sustrik thanks
[08:17] sjampoo aha, you don't need to specify app threads + flags to zmq_init anymore, cool.
[08:20] sustrik yup, I've changed that yesterday
[08:33] CIA-15 zeromq2: 03Martin Sustrik 07master * rf60d891 10/ src/zmq.cpp : Issue 23. zmq_init() crashes on illegal numbers - http://bit.ly/cbVNJE
[08:33] sustrik sjampoo: ok, done
[08:33] sjampoo great.
[08:33] sustrik if you are happy how it behaves now, close the issue
[11:14] mikko morning
[11:38] sustrik morning
[15:01] sjampoo Hmm is zmq_poll in the trunk reliable? I am getting erratic behaviour.
[15:01] sustrik what's happening?
[15:02] sjampoo when i do a zmq_poll only the first item in the set gets it revents updated
[15:02] sjampoo and sometimes not even the first item
[15:03] sustrik hm, do you have a test program?
[15:03] sjampoo yes, http://pastebin.com/yNB5y2J6
[15:03] sjampoo maybe i am doing something wrong
[15:04] sjampoo I tried it with both pub and pair sockets
[15:04] sustrik the code looks ok
[15:04] sustrik it should return ZMQ_POLLOUT for both sockets
[15:04] sjampoo the output is consistently:
[15:04] sjampoo rc: 1
[15:04] sjampoo s:, 2
[15:04] sjampoo r:, 0
[15:05] sjampoo and sometimes: 0,0,0
[15:05] sustrik looks like a bug
[15:05] sjampoo I'll file a report then
[15:05] sustrik yes, please
[15:05] sustrik i'll have a look at it shortly
[15:05] sjampoo i think it might be related to the issue that is already open
[15:06] sustrik which one/
[15:06] sustrik ?
[15:06] sjampoo issue 7 from a month ago, where Poll doesn't return IN events every once in a while
[15:07] sjampoo I think it just doesn't return the first event on both IN and OUT
[15:08] sustrik yes, possible
[15:08] sustrik still, please fill in a ticket
[15:08] sustrik just in case it's a different issue