Re: [netmod] ?==?utf-8?q? [Netconf] YANG library bis model

Hi Rob,

I have forgotten to mention that I have fixed it by modifying the leafref path the exact same way you proposed, it works fine that way.

Regards,
Michal

On Friday, January 26, 2018 10:55 CET, Robert Wilton <rwilton@cisco.com> wrote: 

> Hi Michal,
> 
> Thanks for raising this.
> 
> On 26/01/2018 09:14, Michal Vaško wrote:
> > Hello,
> >
> > we have tried implementing the YANG module ietf-yang-library@2018-01-17.yang from draft-ietf-netconf-rfc7895 and have encountered a problem. I am not completely certain that the issue is with the model and not our XPath evaluator, but based on the definitions I have found I believe the model is wrong.
> >
> >     module: ietf-yang-library
> >       +--ro yang-library
> >          +--ro module-set* [name]
> >             +--ro module* [name]
> >                +--ro name         yang:yang-identifier
> >                +--ro deviation* [module]
> >                   +--ro module    -> ../../name
> >
> > In the tree diagram (I have removed irrelevant parts) it can be seen that there is a "deviation" leafref that should point from one "module" list instance to another identifying the deviation module. So, the supposed evaluation should take the following steps: we start with leaf "module" we take ".." path so we get the "deviation" list. Then we take another ".." and we end up with all the "module" list instances. Finally, we try to find "name" leaf in all these instances of the "module" list.
> >
> > However, our evaluator and some I have found online (all behaving according to XPath definition, I think) proceed differently. At the point of going one step up (parent node) into what es expected to be all the "module" list instances, the context node actually becomes only the *specific module list instance* we have started from, so when looking for the "name" leaf only the name of this list instance is taken into consideration. Needless to say, the evaluation never finds a matching "name" leaf because "module/name" and "module/deviation/module" should never match in one "module" list instance and so the validation always fails.
> It looks like the leafref should be "../../../module/name".
> 
> Does that fix the issue for you?
> 
> Thanks,
> Rob
> 
> 
> > Kind regards,
> > Michal Vasko
> >
> > _______________________________________________
> > Netconf mailing list
> > Netconf@ietf.org
> > https://www.ietf.org/mailman/listinfo/netconf
> > .
> >
> 