Re: [netmod] questions about draft-rtgyangdt-rtgwg-device-model-00

On 08/26/2015 03:48 PM, Andy Bierman wrote:
> 
> 
> On Wed, Aug 26, 2015 at 12:33 PM, Lou Berger <lberger@labn.net
> <mailto:lberger@labn.net>> wrote:
> 
>     Tom,
> 
>     On 8/26/2015 9:34 AM, Nadeau Thomas wrote:
>     > ...
>     >       This is exactly what I want to get on the table.
> 
>     So taking a step back, perhaps there is a YANG language question at the
>     heart of this discussion.  I think we're seeing cases where the same
>     data model is useful in multiple cases/places.  The example I like to
>     use (although I know others disagree with the example) is the case of
>     PE/CE config information, where some of the exact same information may
>     end up on the CE and PE devices as well as the L3 service model.  In
>     this case we'd like a core model to be "included" (or "linked") into two
>     larger models.  Importantly, I'm referring to doing this as part of
>     model definition - not at server/device run time.  This is important for
>     the pre-provisioning case.
> 
>     It is my understanding that there is no way to really do this in a
>     general and extensible way (including allowing for augmentations)
>     today.  If there was such support, I do think we'd be saying that we'd
>     like the existing models to support this mechanism rather than our
>     current proposal of being relocated .
> 
> 
> If you are talking about schema reuse, then YANG has groupings as the
> solution.

My understanding is that the usage scope of groupings is pretty limited
and not really suitable for complex (sub)tree/module representation.
Also groupings can't be augmented

> But it seems you are talking about YANG Mount -- the ability to have a
> subtree on server X represent a different subtree on server Y.  On the
> controller
> the 'mount point' is not the actual data root (as Martin has explained).
> On the NE, the data models are in their real location  On the controller
> they are not.
> 
> This can be done with an 'anyxml' hack today.
> It would be better to have real YANG support for this very basic
> use-case 

I think finding a yang-based solution (to reuse) would be helpful.

Lou

> for YANG Mount.

> 
> 
>  
> 
>     Lou
>     (BTW this is my opinion, not speaking for the DT.)
> 
> 
> 
> Andy
>  
> 
> 
>     _______________________________________________
>     netmod mailing list
>     netmod@ietf.org <mailto:netmod@ietf.org>
>     https://www.ietf.org/mailman/listinfo/netmod
> 
> 