[nanomsg] Re: Routing plan
- From: Schmurfy <schmurfy@xxxxxxxxx>
- To: nanomsg@xxxxxxxxxxxxx
- Date: Thu, 14 Feb 2013 14:38:28 +0100
I agree that devices do not need to be part of the core, they are easy
enough to implement on top of any bindings and would just clutter the core.
On 13 February 2013 17:19, Martin Sustrik <sustrik@xxxxxxxxxx> wrote:
> Hi Chuck,
>
>
> As for executables, it's trivial to just invoke the nn_device function so
>>> it's probably not worth shipping the executables.
>>>
>>
>> I suggest that you reconsider this. In the 0mq community, we have removed
>> and re-added the device functionality 2 or 3 times. The community always
>> demands it back. It might be trivial to implement by user code, but it is
>> quite convenient to have it in the library.
>>
>
> Do you really mean *device executables*? I recall there was such
> adding/removing of zmq_device function, but IIRC there were executables
> once and then they were removed never to return.
>
> The problem with executables is that you have to set them up somehow. So
> you need configuration files (XML, JSON, text?), a parser to read them etc.
> All of that is rather non-standard and would fare better as a separate
> (sub)project IMO.
>
> Martin
>
>
Other related posts:
