[Time] Name | Message |
[00:02] Vagabond
|
anyone aware that the durapub/durasub example doesn't work on git HEAD?
|
[00:03] Vagabond
|
works fine on the latest tarball, but I need the ZMQ_FD sockopt
|
[00:32] iFire
|
does anyone know when Pieter Hintjens is usally on?
|
[12:34] jhawk28
|
I glanced at it
|
[12:35] pieter_hintjens1
|
i guess silence is assent :-)
|
[12:36] pieter_hintjens1
|
anyhow, it means anyone can just submit a signed patch and that's it
|
[12:37] jhawk28
|
that or everyone is sleeping :)
|
[12:37] mato
|
pieter_hintjens1: you might want to announce the new contributions policy on the website/zeromq-announce list as well
|
[12:37] pieter_hintjens1
|
i thought twitter was sufficient :-)
|
[12:38] mato
|
:-/
|
[12:38] pieter_hintjens1
|
Mato: I need to fix the static linking exception
|
[12:38] pieter_hintjens1
|
we agreed to s/iMatix/The authors/?
|
[12:38] mato
|
pieter_hintjens1: let me find the right wording
|
[12:38] mato
|
pieter_hintjens1: hang on a sec
|
[12:38] pieter_hintjens1
|
ack
|
[12:40] mato
|
pieterh: steve has good wording for openpgm
|
[12:40] mato
|
pieterh: http://code.google.com/p/openpgm/source/browse/trunk/openpgm/pgm/LICENSE
|
[12:41] pieterh
|
mato: that's the text we're already using
|
[12:41] pieterh
|
s/iMatix/the copyright holders/
|
[12:41] mato
|
pieterh: yes, but it makes who is granting the exception explicit
|
[12:41] mato
|
pieterh: yeah
|
[12:41] mato
|
pieterh: The copyright holders is IMO best
|
[12:42] pieterh
|
yes
|
[12:42] mato
|
pieterh: Also, the extra paragraph at the end is nice and explicit
|
[12:42] pieterh
|
hmm
|
[12:42] pieterh
|
it's actually not accurate, it's not just a library
|
[12:42] mato
|
well, up to you, you're the expert
|
[12:42] pieterh
|
ackl
|
[12:43] pieterh
|
ack
|
[12:43] mato
|
and in fact, for existing files...
|
[12:43] mato
|
i'd say you probably don't need to update the Copyright statements?
|
[12:50] mato
|
pieterh: you'll want to update this page as well... http://www.zeromq.com/page:join-the-community
|
[12:50] pieterh
|
ack
|
[12:50] mato
|
maybe just put in a short paragraph and refer to the community page
|
[12:51] mato
|
so that we minimize duplication
|
[12:51] pieterh
|
aagh... i had a vpn connection and took it down and now every DNS lookup takes 30 seconds or so...
|
[12:51] mato
|
broken /etc/resolv.conf ?
|
[12:53] pieterh
|
looks broken, how do I fix it?
|
[12:53] mato
|
edit it to contain correct DNS servers?
|
[12:54] mato
|
if you don't know what they are, then somehow force your machine to renew it's DHCP IP
|
[12:54] pieterh
|
argh, it's generated by resolvconf that insists on putting rubbish back into it
|
[12:54] mato
|
ifdown <interface> ; ifup <interface> ?
|
[14:23] mikko
|
sustrik: https://gist.github.com/79a009b8cb51d2d802bc
|
[14:24] mikko
|
occasionally that causes "Assertion failed: term_acks > 0 (own.cpp:175)"
|
[15:24] keffo
|
um.. I have issue with clrzmq
|
[15:25] keffo
|
simplest little rep/req, just passing a string back/forth, and I get "context was terminated", in release
|
[15:25] keffo
|
(and only release)
|
[15:25] keffo
|
100% solid in debug, 100% fail in release.. feels like uninit. variable or something
|
[17:48] ptrb
|
i'm getting 2 messages on my ZMQ_SUB for every 1 I'm sending on the ZMQ_PUB... has this ever come up before, in anyone's memory?
|
[17:49] nisbus
|
hi
|
[17:49] nisbus
|
is the sending speed dictated by the slowest consumer on a pub/sub ??
|
[17:50] ptrb
|
afaik once you have a single subscriber, publish speed is fixed and doesn't depend on subscriber speed
|
[17:51] cremes
|
ptrb: i've never seen that 2 for 1 behavior
|
[17:51] ptrb
|
hmm ok, thanks
|
[17:51] nisbus
|
hmm, I'm getting capped at 2.5 mb/sec
|
[17:52] ptrb
|
what's your hwm?
|
[17:52] nisbus
|
default
|
[17:52] ptrb
|
i might be wrong; cremes or someone would know more
|
[17:52] nisbus
|
if I set it to 1 I should throw away messages that no one is listening to?
|
[17:54] cremes
|
nisbus: yes, a value of 1 will drop
|
[17:55] nisbus
|
great, that's what I'm looking for
|
[17:56] nisbus
|
what about active true/false?
|
[17:56] nisbus
|
what does it mean?
|
[17:57] cremes
|
nisbus: where are you seeing that?
|
[17:58] nisbus
|
erlzmq, the drive for erlang. I've seen examples that look like zmq:socket(pub,[{hwm,1}, {active,true}])
|
[17:59] cremes
|
nisbus: sorry, i don't have an answer
|
[17:59] nisbus
|
np, thanks though
|
[18:00] nisbus
|
I gotta go anyway
|
[18:26] Zelex
|
does ZeroMP support NAT traversal?
|
[18:35] cremes
|
Zelex: yes
|
[18:41] Zelex
|
doesn't there have to be a proxy to keep connections alive in order to do that, and also doesn't it have to be done over UDP (from what I've read, ZeroMQ only supports TCP)
|
[18:42] Zelex
|
basically I'm asking, how does it work, and do I need to do anything special. Connecting between two clients both behind a NAT can get a bit hairy
|
[18:44] PerfDave
|
I don't see that ZeroMQ would need to do anything special to support it, it's just a function of the router.
|
[18:44] Zelex
|
STUN for example
|
[18:45] Zelex
|
its to open a listening port
|
[18:46] PerfDave
|
That's to do with unfriendly routers, though, and again shouldn't be anything to do with ZeroMQ
|
[18:47] Zelex
|
so basically, it doesn't support connections between two clients behind a NAT (without any special port-forwarding setup by the owner of the router)
|
[18:48] Zelex
|
two clients each behind different NATs
|
[18:48] PerfDave
|
why do you think it's its job to do so?
|
[18:49] Zelex
|
cause it makes it infinitely more useful to me
|
[18:49] Zelex
|
:)
|
[18:49] Zelex
|
as in
|
[18:49] Zelex
|
if I can't find a way to do that, I can't use zeroMQ
|
[18:49] PerfDave
|
It doesn't support SLIP either,
|
[18:50] PerfDave
|
But that's nothing to do with ZeroMQ. If you want to develop an application that uses STUN to allow two-way NAT traversal which ZMQ can then use, then go for it.
|
[18:51] Zelex
|
do you know of any project which has done this before?
|
[18:52] PerfDave
|
I mean, I may be talking out of my arse here, but I don't think it's ZMQ's job to fiddle around with packet generation. I'd be surprised and slightly disappointed if it had STUN-type capabilities
|
[18:53] cremes
|
Zelex: i am not aware of any such project
|
[18:53] PerfDave
|
Zelex: Using ZMQ? No. But I'd be surpsised also if you couldn't find a Free STUN library somewhere
|
[18:53] PerfDave
|
Have a look at what Ekiga does perhaps
|
[18:55] Zelex
|
STUN is for UDP right
|
[18:55] Zelex
|
does ZeroMQ support UDP?
|
[18:55] ptrb
|
ZeroMQ sortof replaces UDP
|
[18:56] Zelex
|
yeah, but it fundamentally has to work on top of UDP in order for STUN to work
|
[18:58] Zelex
|
well perhaps I could use STUNT
|
[18:58] Zelex
|
don't know much about it
|
[18:58] Zelex
|
reading up on it now
|
[19:01] Vagabond
|
cool, activity in here
|
[19:04] Vagabond
|
I'm running into some issues with zmq, I need the ZMQ_FD option for getsockopt, but with git HEAD things are kind of strange (the durapub/sub examples don't work, strange crashes, etc)
|
[19:07] cremes
|
Vagabond: do you have a small reproducible example of strangeness with HEAD ?
|
[19:07] Vagabond
|
the mot obvious one is trying the durapub/sub examples from the zmq guide
|
[19:09] Vagabond
|
they work fine with the latest release, but it doesn't work with HEAD
|
[19:09] PerfDave
|
Vagabond: Is there a reason you're not using the latest release?
|
[19:09] PerfDave
|
HEAD sounds like a kind of under development sort of thing :)
|
[19:09] Vagabond
|
I need the ZMQ_FD getsockopt option
|
[19:10] Vagabond
|
I could try to backport that to latest stable
|
[19:10] cremes
|
Vagabond: i recommend that you open a git issue detailing the problem and pointing to the problem code in the guide
|
[19:51] ChicagoBud
|
hello, is pub/sub valid for inproc ?
|
[19:54] Guthur
|
worth a go, I would imagine so at a guess
|
[20:03] cremes
|
ChicagoBud: yes
|
[20:15] ChicagoBud
|
thanks. hit a snag and didn't know if it was supported
|
[20:21] cremes
|
ChicagoBud: it's not supported on Windows, but then nothing using inproc transport works on windows
|
[20:21] ChicagoBud
|
linux
|
[20:21] cremes
|
yeah, no problems on that platform
|
[20:22] cremes
|
make sure you are using release 2.0.9 instead of master unless you need a particular bug fix
|
[20:22] ChicagoBud
|
I think it is because I am sending before the sub are connected
|
[20:25] cremes
|
yes, that's usually the mistake made with pub/sub
|
[20:27] Guthur
|
Is windows likely to get inproc at some point
|
[20:27] Guthur
|
or is it a fundamental limitation of the platform
|
[20:29] cremes
|
Guthur: i don't know... ask sustrik or mato
|
[20:30] cremes
|
.... oops, maybe it's ipc transport that it will never get...
|
[20:30] Guthur
|
I was intending to use inproc, I did read some windows limitation on the guide thing can't remember either
|
[20:30] Guthur
|
I'll check
|
[20:31] Guthur
|
its IPC
|
[20:31] Guthur
|
whew
|
[20:31] Guthur
|
says it may be fixed in future version
|
[20:51] Zelex
|
is there any plans to support UDP protocol?
|
[20:51] Zelex
|
RUDP or the like
|
[20:57] cremes
|
Zelex: hack the code; according to the imatix guys, adding a new protocol is pretty easy and already abstracted away from the core library
|
[20:58] Zelex
|
alright, I'll take a look at whats involved there
|
[21:03] starkdg
|
hello !
|
[22:43] Zelex
|
does Ekiga use ZeroMQ?
|