interesting, but unpersuasive, talk.
i think lightning talks are poorly suited to explaining why zeromq is the bomb.
there are few things i would like to do more than to
get together with pieter and hammer out how to effectively communicate
why zeromq is really a big part of teh answer of how to write distributed
programs.
(pieter do you ever swing by the usa? or can we video conference somehow?)
the main issue is that of mindset, i think. writing distributed programs is the
hardest thing programmers do. we seem poorly suited to reason about processes
running in parallel, and the normal tools on mutexes et al are fragile and hard
to use.
accordingly, i have actively avoided distributed programming for over 25 years.
with one interesting exception: using the ALEF language by phil winterbottom
running on brazil (the OS between plan9 and inferno at bell labs research).
here the ONLY ipc i used were buffered channels (a la Hoare), and the language
supported threads and processes. it is ironic that 20 years later, i find myself
with nearly the same tools again, and once again i am writing distributed
programs.
to get back to my point, eschewing pointers and mutexes seems to be the key;
use channels (err, zmq sockets) for single threading (err, critical regions)
when needed.
let me leave with another thought. cfengine (mark burgess, university of oslo)
has
been the dominant system configuration tool for many years. a few years back,
mark realised that the notion of a central node ordering around a gaggle of
nodes
to achieve a specific result was wrong, and in the coming era of unreliable
nodes,
unworkable. he has changed the fundamental notion to that of nodes
self-organising
via "promises" (basically node-to-node contracts about performed services).
this mindset change has been hard to stomach for many cfengine users, and it
strikes
me as quite analogous to the change that zeromq demands of its users.
andrew
On Feb 6, 2011, at 1:29 AM, Pieter Hintjens wrote:
> Hi All,
>
> The video of my breathless lightning talk on 0MQ is now up here:
> http://streaming.fosdem.org:8000/ferrer/0mq.ogv
>
> And the slides are here: http://www.slideshare.net/pieterh/fosdem-2011-0mq
>
> Hopefully this approach to "What is 0MQ and why would I want to look
> at it" can help spread the word. :-)
>
> -
> Pieter Hintjens
> iMatix
> _______________________________________________
> zeromq-dev mailing list
> [email protected]
> http://lists.zeromq.org/mailman/listinfo/zeromq-dev
------------------
Andrew Hume (best -> Telework) +1 623-551-2845
[email protected] (Work) +1 none currently
AT&T Labs - Research; member of USENIX and LOPSA
_______________________________________________
zeromq-dev mailing list
[email protected]
http://lists.zeromq.org/mailman/listinfo/zeromq-dev
