Re: [YANG] new pyang errors

On Wed, Jan 23, 2008 at 01:02:43PM +0100, Ladislav Lhotka wrote:

> > But we seem to have fundamental differences about the split of work:
> > 
> > - I want the NETCONF server to be authoritative to tell the client what
> >   the configuration is, with and without defaults expanded.
> 
> A simpler protocol is better. If you allow this choice, you also have
> to:
> 1. extend the NETCONF protocol so that it is able to distinguish 
>    between the two cases,
> 2. carefully specify the semantics of concepts like default values, 
>    mandatory versus optional etc.

I think this is why we are discussing these questions here.

> > - You seem to assume some magic processing exists on the manager side
> >   which takes what the device reports and modifies it e.g. by patching
> >   in defaults yielding a complete config.
> 
> Why do you call it magic? It's perfectly deterministic and local
> processing from client's point of view - the application has all
> information and just does what the user wants. And the user can always
> check what's going on without caring about what "magic" happened at the
> remote end. It can efficiently work even with different devices - the
> factory defaults are not relevant here.

The point is that the application does not really kow whether it has
all the information and it is a pain for the user to check this.

> It does imply slightly more complexity than raw ssh on the manager side,
> but for the managed device it is actually simpler - it needn't care
> about which values in its config datastore differ from factory defaults.

Lets agree to disagree on this. We won't convince each other.

/js

-- 
Juergen Schoenwaelder           Jacobs University Bremen gGmbH
Phone: +49 421 200 3587         Campus Ring 1, 28759 Bremen, Germany
Fax:   +49 421 200 3103         <http://www.jacobs-university.de/>

_______________________________________________
YANG mailing list
YANG@ietf.org
https://www1.ietf.org/mailman/listinfo/yang