On 14 October 2010 13:23, Pieter Hintjens wrote:
> Ack on your points. KeepAlive / Heartbeating belongs in the tcp://
> transport layer, doing this at the application level does not appear
> to be the best approach. But it's the workaround for now. You can
> make it work better by only sending
Martin,
Thanks! The patch is applied to master.
Martin
On 10/14/2010 08:24 AM, Martin Pales wrote:
Hello,
the attached patch corrects a minor typo in devpoll_t::loop to resolve
compilation error in Solaris.
Martin
___
zeromq-dev mailing list
ze
On 10/14/2010 07:18 AM, Pieter Hintjens wrote:
> Oliver,
>
> On Thu, Oct 14, 2010 at 1:49 AM, Oliver Smith wrote:
>
>> Recently upgraded from an earlier 2.x (2.0.7 I think) to 2.1.0. I
>> immediately ran into an unusual problem under Linux where my unit test
>> module passed all tests but th
Oliver,
>Recently upgraded from an earlier 2.x (2.0.7 I think) to 2.1.0. I
> immediately ran into an unusual problem under Linux where my unit test
> module passed all tests but then failed to terminate.
>
> The test unit keeps getting stuck with this backtrace, apparently in the
> destructor f
Hello,
the attached patch corrects a minor typo in devpoll_t::loop to resolve
compilation error in Solaris.
Martin
0001-zmq-devpoll_t-correct-a-typo-in-loop.patch
Description: Binary data
___
zeromq-dev mailing list
zeromq-dev@lists.zeromq.org
http://
Ack on your points. KeepAlive / Heartbeating belongs in the tcp://
transport layer, doing this at the application level does not appear
to be the best approach. But it's the workaround for now. You can
make it work better by only sending heartbeats when there is no other
data to send, i.e. when
Oliver,
On Thu, Oct 14, 2010 at 1:49 AM, Oliver Smith wrote:
> Recently upgraded from an earlier 2.x (2.0.7 I think) to 2.1.0. I
> immediately ran into an unusual problem under Linux where my unit test
> module passed all tests but then failed to terminate.
>
> The test unit keeps getting stuck
On Wed, 13 Oct 2010 at 5:13 PM, Pieter Hintjens wrote:
> On Wed, Oct 13, 2010 at 4:26 PM, Martin Lucina wrote:
>
> > In Marcelo's specific case he will need to add it into other traffic;
> > adding an extra socket won't help since his network will still declare
> the
> > normal idle connection d
Recently upgraded from an earlier 2.x (2.0.7 I think) to 2.1.0. I
immediately ran into an unusual problem under Linux where my unit test
module passed all tests but then failed to terminate.
The test unit keeps getting stuck with this backtrace, apparently in the
destructor for a static zmq::
Hi Guys,
I've updated stack_context.py and zmqstream.py to incorporate recent
changes from Tornado.
http://github.com/zeromq/pyzmq/pull/38
In addition to fixing some other bugs we found in Tornado, this
appears to fix the issue where close() was hanging. I didn't undo the
DelayedCallback logic y
Yeah, makes sense.
- mh
On Wed, Oct 13, 2010 at 10:15 PM, Martin Sustrik wrote:
> On 10/13/2010 10:11 PM, Martin Hurton wrote:
>>
>> Martin, any reason to postpone the initialization of the end variable?
>
> It's an attempt to lower the latency. If there's already an event to return
> when you i
On 10/13/2010 10:11 PM, Martin Hurton wrote:
> Martin, any reason to postpone the initialization of the end variable?
It's an attempt to lower the latency. If there's already an event to
return when you invoke zmq_poll, the call to gettimeofday is skipped.
Martin
Martin, any reason to postpone the initialization of the end variable?
- mh
On Wed, Oct 13, 2010 at 9:43 PM, Martin Sustrik wrote:
> Martin & Martin,
>
> Thanks for comments! Here's try no.2
>
> Martin
>
___
zeromq-dev mailing list
zeromq-dev@lists.zer
Thanks, I've added it to the collection!
On Wed, Oct 13, 2010 at 9:48 PM, Dogen Punk wrote:
>
> I noticed this hadn't been done yet...
>
> "You can't be idealistic in this world and not be crazy. Because they've
> created such a deep structure now, you can't get in. And we don't want to
> get in,
I noticed this hadn't been done yet...
"You can't be idealistic in this world and not be crazy. Because they've
created such a deep structure now, you can't get in. And we don't want to
get in, we're on the outside. But we're not on the outside looking in, we're
on the outside looking out. So I fe
Martin & Martin,
Thanks for comments! Here's try no.2
Martin
>From e2167cecaefec6557c7a5712fb75e51487ff69a6 Mon Sep 17 00:00:00 2001
From: Martin Sustrik
Date: Wed, 13 Oct 2010 21:39:20 +0200
Subject: [PATCH] Precise timouts in zmq_poll implemented
Signed-off-by: Martin Sustrik
---
src/zmq.c
2010/10/13 Pieter Hintjens :
>> Hi, as I can read in that page the element doing the failover is a
>> queue rather than the client itself, am I right? Does it mean that a
>> single client (REQ socket) couldn't not use this failover feature by
>> itself?
>
> Sure it could, it just needs to implement
It seems the zmq_poll can now wait infinitely long when timeout is 0.
- Martin
On Wed, Oct 13, 2010 at 1:59 PM, Martin Sustrik wrote:
> Hi all,
>
> Here's a patch to master that should prevent zmq_poll exiting when specified
> timeout is not yet reached.
>
> Testing and reporting problems is wel
On Wed, Oct 13, 2010 at 5:18 PM, Iñaki Baz Castillo wrote:
> 2010/10/13 Pieter Hintjens :
>> The code isn't finished but here is a design sketch:
>>
>> http://github.com/imatix/zguide/blob/master/articles/reliability.md
> Hi, as I can read in that page the element doing the failover is a
> queue
On Wed, Oct 13, 2010 at 4:26 PM, Martin Lucina wrote:
> In Marcelo's specific case he will need to add it into other traffic;
> adding an extra socket won't help since his network will still declare the
> normal idle connection dead, so the hearbeat (or in this case keepalive)
> needs to travel a
OK, I'll ask github if they can do something...
On Wed, Oct 13, 2010 at 6:04 PM, MinRK wrote:
> You can do it well after the fact. We only did it last week. All it does
> is change how it's viewed on GitHub, but it's nicer for the main repo to not
> say 'forked from ...'.
> -MinRK
>
> On Wed, O
You can do it well after the fact. We only did it last week. All it does
is change how it's viewed on GitHub, but it's nicer for the main repo to not
say 'forked from ...'.
-MinRK
On Wed, Oct 13, 2010 at 02:52, Pieter Hintjens wrote:
> Hey Benjamin,
>
> Thanks for this pointer, but I'd alread
I rewrite patch file and add "Signed-Off-By" tag. Sorry if I do
something wrong - I newbie here.
--
Oleg Sevostyanov
JID: boo...@jabber.ru
>From 89eeb3af58b4095e47c68c84a4b569fd6a6858dc Mon Sep 17 00:00:00 2001
From: Oleg Sevostyanov
Date: Wed, 13 Oct 2010 09:18:18 +0400
Subject: [PATCH] add
2010/10/13 Pieter Hintjens :
> The code isn't finished but here is a design sketch:
>
> http://github.com/imatix/zguide/blob/master/articles/reliability.md
>
> There are some issues making this work simply, mainly that zmq_poll
> has the 'feature' of returning unpredictably. There's a workaround
>
Hi guys,
I may also attend; will see how things work out next week.
-mato
___
zeromq-dev mailing list
zeromq-dev@lists.zeromq.org
http://lists.zeromq.org/mailman/listinfo/zeromq-dev
Martin,
sust...@moloch.sk said:
> Here's a patch to master that should prevent zmq_poll exiting when
> specified timeout is not yet reached.
Looks good. I've not actually tested the code, just read it. Suggest you do
the select() implementation also.
If I read this correctly the semantics chan
2010/10/13 Pieter Hintjens :
> The code isn't finished but here is a design sketch:
>
> http://github.com/imatix/zguide/blob/master/articles/reliability.md
>
> There are some issues making this work simply, mainly that zmq_poll
> has the 'feature' of returning unpredictably. There's a workaround
>
p...@imatix.com said:
> You're hitting an issue that most 0MQ users come to sooner or later.
> 0MQ sockets do not, currently, do any kind of heartbeating over TCP
> though this is something I'm pretty sure they eventually will need to
> do.
>
> The solution is fairly simple, add heartbeating to wh
sust...@moloch.sk said:
> Hi Oleg,
> > I try to build last git in MSVC 2005 but compiler not found __rdtsc ()
> > fucntion. I make changes in clock.cpp and send patch file.
> b
> Two issues:
>
> 1. I believe _MSC_VER should be used instead of ZMQ_HAVE_WINDOWS to keep
> the MinGW build OK. (Not s
Hi all,
Here's a patch to master that should prevent zmq_poll exiting when
specified timeout is not yet reached.
Testing and reporting problems is welcome!
Martin
>From 9d1306d6bbee66ad2285d164a37df2f6d1dde741 Mon Sep 17 00:00:00 2001
From: Martin Sustrik
Date: Wed, 13 Oct 2010 13:58:52 +020
Hi all,
Two small news items.
1, we're going to organize a 0MQ conference early next year. Details
to come later.
2, we're launching a competition to find as many ways to crash 0MQ as
possible by sending it invalid messages. Winner gets a free trip to
the conference.
Details on http://www.zer
Hi Delaney,
So, if 0MQ asserts or crashes when passed an invalid message on the
wire, we consider this a bug. Since you have a nice test case, could
you report this to the 0MQ issue tracker, and post your code to a gist
(gits.github.com)?
-Pieter
On Wed, Oct 13, 2010 at 12:03 PM, Pieter Hintjen
Hi Marcelo,
You're hitting an issue that most 0MQ users come to sooner or later.
0MQ sockets do not, currently, do any kind of heartbeating over TCP
though this is something I'm pretty sure they eventually will need to
do.
The solution is fairly simple, add heartbeating to whatever protocol
you a
On Wed, Oct 13, 2010 at 12:46 AM, Delaney Gillilan
wrote:
> Just starting out with pyzmq and zeromq in general. So far the guides been
> really helpful. There is no example of how to hook up a REQ/XREP connection
> using polling. I'm sure its something simple. I've tried
> socket.recv_multipa
Hey Benjamin,
Thanks for this pointer, but I'd already done it the old way, clone
and wipe the old repository. I guess I could backtrack now but more
than likely I'll make a mess of it. :-)
-Pieter
On Tue, Oct 12, 2010 at 8:15 PM, MinRK wrote:
> If you want that to be the new parent as seen o
On Tue, Oct 12, 2010 at 4:55 PM, Iñaki Baz Castillo wrote:
>> I'll make a worked example for the Guide, it'll take me a couple of days.
>
> Hi Pieter, any new about it? is it planned for some specific date? :)
The code isn't finished but here is a design sketch:
http://github.com/imatix/zguide/
36 matches
Mail list logo