Re: [netconf] Generic Capabilities model

Hi,

I strongly support this generic capabilities work.
IMO this approach would make NMDA much easier for a client to support,
since the new /yang-library
would not be needed for this purpose.  Instead, a simple "nmda-operational"
capability could be used in this module
to easily identify which config=true nodes should be expected in
<operational>.

The complexity in /yang-library might be justified if new writable
datastores (like <ephemeral>) are
ever introduced, but not for solving the problem "what features does this
server support for the specified
data node (resource)"?

Andy

On Thu, Dec 5, 2019 at 9:15 AM Benoit Claise <bclaise@cisco.com> wrote:

> Hi Qin,
>
> I would agree.
> I believe it's important to have a generic solution. I can picture more
> augmentations.
> The change is trivial IMO, i.e. modify the ietf-notification-capabilities
> model
>
>    - Rename the top level container to system-capabilities
>    - Add a container subscription-capabilities to the grouping
>    subscription-capabilities to contain all subscription related capabilities
>    - Invite others to augment the model with similar groupings for other
>    capabilities.
>
> Regards, Benoit.
>
> Hi, Balazs:
>
> In last IETF meeting, you offered a proposal in netmod session on
> notification capability change that was discussed in netmod session.
>
> I think it is a good idea to define generic capabilities model in
> draft-ietf-netconf-notification-capabilities
>
>
> https://datatracker.ietf.org/meeting/106/materials/slides-106-netmod-sessb-generic-model-for-server-capabilities-00
>
> since we have other capabilities that need to be covered, one of example
> such capability is one that can be self-described in
>
> draft-tao-netmod-yang-node-tags
> <https://tools.ietf.org/id/draft-tao-netmod-yang-node-tags-00.txt>.
>
> With such new capability, should we augment from YANG Push model or should
> we augment from notification capability?
>
> We see one downside of augmenting from YANG Push model, is it only can be
> used in the running time, it can not be used in the design time or
>
> Implementation time.
>
>
>
> So I think if one generic capability model can be defined, it will allow
> more flexibility to add new capability. However if we decide to take this
> approach,
>
> Probably notification capabilities draft require substantial changes to
> the current model structure. But I think it worth to do so, in my personal
> view.
>
>
>
> -Qin
>
> _______________________________________________
> netconf mailing listnetconf@ietf.orghttps://www.ietf.org/mailman/listinfo/netconf
>
>
> _______________________________________________
> netconf mailing list
> netconf@ietf.org
> https://www.ietf.org/mailman/listinfo/netconf
>