Re: [Netconf] [yang-doctors] YANG Doctor question: empty mandatory choice? (was RE: YangPush now)

["Eric Voit (evoit)" <[e-mail]>]

> From: Kent Watsen, Friday, July 27, 2018 2:03 PM
> 
> > I would prefer to not add empty mandatory choices to any YANG module.
> > An empty choice is bad enough. In 2119 terms, using this container is
> > a SHOULD, not a MUST.
> 
> Moving past the syntax-level discussion:
> 
> 1) do you agree that a transport MUST be configured?
> 
>        - if not, then how is this a configuration model when half the config is
>          hidden behind vendor magic?  In my view, this pushes the limit of
>          reasonableness.

I think transport MUST be configured.   

But as today's transport configuration is vendor specific, I am not sure as to whether this is a requirement which the SN YANG model structure itself enforces.  (See next comment.)

>   2) do you agree that at most only one transport can be configured per
> "receiver"?

Yes.  

There are alternative ways to do this.  For example we could assert that when a "-notif" draft designer does a augmentation of transport specific objects, those augmentations MUST include XPATH constraint from those objects to a specific instance of "transport" under the subscription.   (E.g., you can only augment leafref instances pointing to ietf-netconf-server if the subscription's transport is "NETCONF".)  For example:


module ietf-netconf-subscribed-notifications {
 
  prefix nsn;
 
  import ietf-netconf-client { prefix ncc; }
  import ietf-subscribed-notifications { prefix sn; }

  identity netconf {
    base sn:transport;
    base sn:inline-address;
    description
      "NETCONF is used as a transport for notification messages and
       state change notifications.";
  }

  augment "/sn:subscriptions/sn:subscription/sn:receivers/sn:receiver" {
   when 'derived-from(../../../transport, "nsn:netconf")';   
   description
      "This augmentation allows NETCONF specific parameters to be exposed for a receiver.";
    leaf netconf-endpoint {
      type leafref {
        path "/ncc:netconf-client/ncc:initiate/ncc:netconf-server/ncc:endpoints/ncc:endpoint/ncc:name";
      }
      description
        "Remote client which need to initiate the NETCONF transport if an existing NETCONF session from that client is not available.";
    }
  }
  
}

With such models, one transport per receiver should result.

Eric

>         - this is perhaps arguable but, if needed, then I'd push for a list with
>           min-elements 1.   That said, why support multiple transports under
>           a single "receiver", when multiple "receiver" instances can be created,
>           one for each?   Additionally, it muddies what the text means by
>           "receiver state", as then there would have to be state per transport.
> 
> Kent // contributor