ZeroMq IRC Log

Saturday February 5, 2011

[Time] Name	Message
[00:13] Steve-o	but how do you want the timeout handled?
[00:16] zchrish	Well, I will implement a handler of course. I am just trying to figure out how to configure 0mq to implement a timeout.
[01:10] Leolo_3	zchrish : zmq_poll() with a timeout value
[01:25] zchrish	I see; thank you.
[09:52] jugg	running ./autogen.sh using autoconf 2.67 (as installed from debian 6), produces this: configure.in:9: warning: AC_INIT: not a literal: m4_esyscmd([./version.sh \| tr -d '\n']) And this patch fixes it: http://paste2.org/p/1229684 - Will this cause issues on other systems? If not I'll go through the steps of submitting the patch.
[10:50] Guthur	Has the reason been identified why when two sockets connect with the same identity it crashes the destination?
[11:13] sustrik	jugg: you have to speak to mikko
[11:13] sustrik	he's in charge of the build system
[11:13] sustrik	Guthur: i assume there's an assertion there
[11:13] sustrik	not sure
[11:22] Guthur	right, assert(_engine)
[11:22] Guthur	or engine_ rather, let me find the line
[11:22] Guthur	it was on that stack trace I posted yesterday
[11:25] Guthur	I personally feel it should be reasonably high priority bug, having clients being able to crash the server so easily is not good
[11:26] Guthur	and from what I have experienced the server app will not even receive reasonable feedback on the error
[11:26] sustrik	yup, feel free to fix it
[11:29] Guthur	the assert in question is zmq_assert (engine_) in zmq::session_t::process_attach
[11:34] Guthur	peer_identity_ would be the identity, correct?
[11:37] sustrik	peer_identity_ is the identity of the peer socket
[11:51] Guthur	I'm going to slot this on to my todo list, but lack of inner zeromq knowledge will make it hard to find
[11:52] Guthur	I want to add a couple of things to clrzmq2 today
[11:54] Guthur	I was actually thinking of adding a clrzmq-extensions project, with a special form of queue device, and generic send/recv with .Net serialization
[12:05] Guthur	Umm, pub/sub using inproc, seems to have stop working for me, very odd
[12:07] Guthur	actually scratch that, its inproc in general
[12:11] sustrik	inproc stopped working?
[12:13] Guthur	must be something to do with this computer
[12:13] Guthur	it was working fine in work
[12:13] Guthur	and it worked fine on this before
[12:14] Guthur	the peer has connection refused
[12:23] Guthur	ah got it
[12:23] Guthur	I did not realize inproc connection had to be the same context
[12:23] Guthur	learnt something new
[12:32] sustrik	:)
[12:52] mikko	good morning
[13:00] jugg	mikko, I just sent a patch to the mailing list, making a change to configure.in - I don't know if it'll have any side affects on other systems.
[13:07] mikko	let me check, sec
[13:43] CIA-21	zeromq2: 03Chris Rempel 07master * r4efead1 10/ (AUTHORS configure.in):
[13:43] CIA-21	zeromq2: Fixed autogen.sh warnings on Debian 6 systems.
[13:43] CIA-21	zeromq2: Signed-off-by: Chris Rempel <csrl@gmx.com>
[13:43] CIA-21	zeromq2: Please enter the commit message for your changes. Lines starting - http://bit.ly/dHdxHE
[13:50] jugg	sustrik, is there any point in opening pull requests on github for zeromq?
[13:53] sustrik	jugg: no
[13:53] jugg	ok
[13:53] sustrik	the process works via the mailing list
[14:02] jugg	tests/test_hwm should be added to .gitignore
[14:04] sustrik	jugg: let me see
[14:05] CIA-21	zeromq2: 03Martin Sustrik 07master * rc8e5be8 10/ .gitignore :
[14:05] CIA-21	zeromq2: test-hwm added to .gitignore
[14:05] CIA-21	zeromq2: Signed-off-by: Martin Sustrik <sustrik@250bpm.com> - http://bit.ly/hLOv45
[14:05] sustrik	done
[15:14] mikko	does anyone know if Neale is on IRC?
[15:17] sustrik	mikko: he was here few day ago
[15:17] sustrik	but i haven't seen him since
[15:24] mikko	mailed him
[15:24] mikko	it would be very nice to get system z
[15:24] mikko	and maybe solaris sparc at some point
[15:31] sustrik	i think mato has a solaris box in the cellar
[15:31] sustrik	you may ask him whether it's possible to put it online after's he back from NZ
[15:40] Guthur	ah the pain of maintaining backwards compatibility
[15:40] Guthur	though the problem is my own doing
[15:41] Guthur	I should have made the Poll method static in clrzmq2
[15:48] sustrik	you mean the original problem you've reported?
[16:08] Guthur	oh no sorry, unrelated
[16:08] Guthur	just lamenting
[17:34] traviscline	Any idea on the expected amount of overhead associated with using an external event loop ala ZMQ_FD?
[18:05] Guthur	Would it be ok to put code in the clrzmq2 repo that should be considered beta
[18:06] Guthur	It doesn't effect the binding, it's just extension/helper stuff
[18:10] Leolo_3	travis : I suspect very very little. ZMQ does it's stuff in a background thread, and sends messages over the FD. It does this "normally" when you call zmq_recv().
[18:20] traviscline	Leolo_3: thanks