Re: [YANG] insert and key attributes

Phil Shafer wrote:
> Andy Bierman writes:
>> Conformance to RFC 4741 does not require any additional
>> attributes for <edit-config>.
> 
> Nor does it prevent the use of additional attributes.
> 
>> If the same operation
>> without the YANG attributes does not produce the same result,
>> like the 'move barney' example, then the agent is not RFC 4741
>> compliant.
> 
> Operations on NETCONF data models are defined by the data model.
> The data model is the contract between the agent and the manager,
> and if neither party understands this contract, operations are not
> possible.  NETCONF doesn't define a generic data model that all
> parties must implement or understand.
> 
> There is no requirement that I support the xc:operation attribute
> on every (or any) node in my data model.  NETCONF gives lots of
> freedom to the data modelers.  This is by design.
> 
> If my data model defines additional operations, attributes, or
> elements, this doesn't invalidate any claim of RFC 4741 conformance.
> 
> YANG is a means of defining data models which follow a particular
> set of rules.  One of these rules is this move operation.  In order
> to implement a YANG-based data model, an agent will need to support
> this operation.  But this will not make that agent non-compliant
> with 4741.
> 

This is your interpretation of RFC 4741.
I think a WG should decide matters of standards conformance.

I agree with Balazs that changes to protocol behavior belong
in a different document, and should not be part of the data
modeling language.

I do not think it promotes multi-vendor interoperability
to have each vendor come up with their own attribute extensions
to NETCONF <edit-config>.  The goal of the IETF should be to
converge on a standard <edit-config> operation, not simply
codify proprietary CLI with angle brackets.

> Thanks,
>  Phil
> 
> 

Andy

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