[zeromq-dev] czmq / zauth

Greg Young gregoryyoung1 at gmail.com
Fri May 6 22:00:00 CEST 2016


As we discussed in Vilnius there are some disadvantages to the
thread/actor route. In particular the creation of dynamic actors
becomes quite expensive (eg actor/request as example). Often in such
circumstances a developer is forced to reimplement much of the logic
of the model in order to keep their static layout (eg a hashtable of
request correlation ids on a single actor). The thread/actor model is
quite well suited to static layouts but runs into issues with dynamic
layouts. Are dynamic layouts a thing people may be interested in is
also an interesting discussion.

On Fri, May 6, 2016 at 10:54 PM, Pieter Hintjens <ph at imatix.com> wrote:
> We did a lot of work with green threads in the past, in C. In fact we
> kind of do the same with our zproto state machine, where each
> connection is a "thread". In other places it hardly seems worth it.
> System threads are simple and effective for our usage of actors.
>
> On Fri, May 6, 2016 at 9:34 PM, Greg Young <gregoryyoung1 at gmail.com> wrote:
>> I am curious if anyone would be interested in trying to quantify the
>> work involved with green threads on zmq actors?
>>
>> On Fri, May 6, 2016 at 10:29 PM, Pieter Hintjens <ph at imatix.com> wrote:
>>> The current actor API is a bit simplistic. I'd have liked to redesign
>>> it using the new client/server sockets so that you could talk to
>>> actors from any thread. As it is, there's no explicit mechanism to
>>> check if actor X is running or not. You have to start the actor in
>>> your main thread. You might be able to detect zauth itself by trying a
>>> connect that needs it (as we do in the security selftests).
>>>
>>> On Fri, May 6, 2016 at 8:17 PM, Brian Knox <bknox at digitalocean.com> wrote:
>>>> Is there currently a way one can check if zauth is running for a process -
>>>> e.g.,  imagine a case where a zauth actor may be been started by another
>>>> thread - is there a way one can check (in a thread safe manner) if this was
>>>> the case?
>>>>
>>>> Brian
>>>>
>>>> _______________________________________________
>>>> zeromq-dev mailing list
>>>> zeromq-dev at lists.zeromq.org
>>>> http://lists.zeromq.org/mailman/listinfo/zeromq-dev
>>> _______________________________________________
>>> zeromq-dev mailing list
>>> zeromq-dev at lists.zeromq.org
>>> http://lists.zeromq.org/mailman/listinfo/zeromq-dev
>>
>>
>>
>> --
>> Studying for the Turing test
>> _______________________________________________
>> zeromq-dev mailing list
>> zeromq-dev at lists.zeromq.org
>> http://lists.zeromq.org/mailman/listinfo/zeromq-dev
> _______________________________________________
> zeromq-dev mailing list
> zeromq-dev at lists.zeromq.org
> http://lists.zeromq.org/mailman/listinfo/zeromq-dev



-- 
Studying for the Turing test



More information about the zeromq-dev mailing list