[IPython-dev] Zmq branch on LP: re-created.

Brian Granger ellisonbg@gmail....
Thu Apr 29 11:28:34 CDT 2010


Hans,

0MQ has very smart reconnection logic, and we take advantage of that
in the huge way.  For example, you can send messages to a process that
is not currently running.  Those message will be queued and when the
process appears, 0MQ detects that and will send the messages.  Also,
0MQ will disconnect/reconnect its raw sockets many times if needed.
The bottom line is that you can't detect the types of failures you are
talking about at the user API level.  At first, this bothered me as I
am used to handling the disconnect by hand.  But, as time has gone on,
I see the huge advantages this approach gives.  In this new way of
thinking a heartbeat is the only way to detect if a peer is truly
alive.

Cheers,

Brian

On Thu, Apr 29, 2010 at 12:50 AM, Hans Meine <hans_meine@gmx.net> wrote:
> Am Dienstag 27 April 2010 07:32:16 schrieb Fernando Perez:
>> > 2) is good idea make a daemon to crash recovery system?
>>
>> Brian and I thought a bit about this, and we think that it might be
>> possible to run a 'heartbeat' socket directly in zmq.  But I  wouldn't
>> worry about this too much yet.
>
> In our own two-process, Qt-based python terminal, we simply connected to the
> signal the TCP socket class emits when the connection closed.  AFAIK that's
> the standard approach for TCP/IP-based IPC with daemons (e.g. I recall
> KDE/DCOP etc. do that), and if 0mq really is the superior solution nowadays,
> I hope it offers the same failsafe detection of a lost/closed connection.
>
> HTH,
>  Hans
> _______________________________________________
> IPython-dev mailing list
> IPython-dev@scipy.org
> http://mail.scipy.org/mailman/listinfo/ipython-dev
>



-- 
Brian E. Granger, Ph.D.
Assistant Professor of Physics
Cal Poly State University, San Luis Obispo
bgranger@calpoly.edu
ellisonbg@gmail.com


More information about the IPython-dev mailing list