Re: [netconf] [netconf-wg/https-notif] Should the receivers container be moved under the augment statement and the leafref renamed? (#2)

[Kent Watsen <[e-mail]>]

Hi Martin,

>>>> The reason to not move the receiver container under the augment is so
>>>> as to allow the leafref to point to multiple receivers.
>>> 
>>> I don't understand this reason.
>> 
>> Each "receiver" (in https-notif) maps to an HTTPS connection from the
>> publisher to the receiver.
>> 
>> There is likely to be more than one configured subscription, yet all
>> notifications should go to the same receiver.
>> 
>> We'd like to use the same HTTPS "connection" for all, as opposed to
>> having an HTTPS connection for each.
>> 
>> The "receiver-ref" leaf provides an indirection enabling this
>> many-to-one relationship.
> 
> I can't find the original thread about this issue, so I'll rephrase
> what I (think I) wrote from the start:
> 
> I understand the reason for the many-to-one, and I wish we had that in
> the base model itself (i.e., RFC 8639), so that protocol-documents
> didn't have to invent this, and so that we could have
> receiver-specific config in one place.

I agree, it would’ve been better in RFC 8639.  I recall Juergen writing once that he expected it to be a common pattern.  But that is behind us now, or are you suggesting a -bis?  Is there is an action coming out of this fork in the thread?


>>> Since this is not a stand-alone model, I think it should augment
>>> /sn:subscriptions.  In some way it doesn't matter what nodes are
>>> called and where they are located, but having descriptive names and
>>> keep related nodes under common subtrees helps the understanding of
>>> models.
>> 
>> Is s/receiver-ref/https-receiver-ref/ what you had in mind?
> 
> Yes, as a minimum.
> 
> I would also change the top-level container "receivers" to augment
> /sn:subscriptions:
> 
>  augemnt /sn:subscriptions {
>    container https-receivers {
>      ...
>    }
>  }

Ah, I see now, thank you for providing the snippet.


Kent // contributor