GSoC 2008: netconf suggesion
Andreas Louca
alouca at neutrix.net
Thu Mar 27 11:47:57 UTC 2008
I totally agree with you.
The problem is I don't really understand whats the ultimate target.
So lets recap, we want something that just works and completed fast in
order to provide some functionality up until a milestone of a first
release and get some feedback back to the project, after being
accepted in Debian lothy in August. At a later time (out of the scope
of GSoC), the netconf would be rewritten, in C/C++, laying down some
more extensive foundations, in order for netconf to be more
"mainstream".
On Thu, Mar 27, 2008 at 1:08 PM, Colin Alston <colin at thusa.co.za> wrote:
> On 27/03/2008 12:57 Andreas Louca wrote:
> > What I meant it doesn't take a long time to define a few API calls,
> > that would be a guideline for later.
>
> Just to interject on that point.
>
> It may seem "simple" to define a bunch of API calls but you have to
> remember the whole point of an API. It's not something that can be
> micro-managed well in parallel with development unless you have a very
> very well defined structure.
>
> The thing of primary importance is that your API components should
> NEVER change. It can gain functionality but if you discover a function
> should require more parameters or some such later on in the project,
> if that API change causes an architectural shift then you have a
> catastrophic problem.
>
> So still I feel even a small Bridge API is not something to be taken
> lightly. Considering the current netconf strategy is not necessarily a
> throw-away prototype, you still need a plan and it needs to be well
> executed - not rushed.
>
> Regards,
> --
> Colin Alston <colin at thusa.co.za>
> System Analyst, Linux & Internet Services
> Thusa Business Support (Pty) Ltd
>
> "Usually the protocol is this: I appoint someone for a task,
> which they are not qualified to do. Then, they have to fight
> a bear if they don't want to do it." -- Glyph Lefkowitz
>
>
More information about the netconf-devel
mailing list