[core] Re: Additional CORECONF Comments

[Andy Bierman <[e-mail]>]

On Thu, Jan 15, 2026 at 1:31 PM Andy Bierman <[e-mail]> wrote:
>
> Hi,
>
> On Thu, Jan 15, 2026 at 9:48 AM Vojtech Vilimek
> <vojtech.vilimek=[e-mail]> wrote:
> >
> > Hi all,
> >
> > I found a crucial bug in the CORECONF: You can discover all the schema
> > nodes of given model in the yang-library or .well-known/core but how do
> > you discover instances of a list? For example in the following YANG model.
> >
>
>
> I brought this issue up as well:
> https://mailarchive.ietf.org/arch/msg/core/jIIuha0fonaDvjtYc9iFcKo5BgI/
>

Sorry this is the wrong link.
The message I meant is dated  March 17, 2024, and has subject:

Re: [core] WG Last Call on draft-ietf-core-comi-16

It resulted in a github issue:
https://github.com/core-wg/comi/issues/15


Andy



Andy

> The RESTCONF 'fields' solution is more general and fairly complex.
> The 'keys-only' boolean proposal should be easy to implement and
> encode in a request.
>
> I think you are right that for operation within a constrained network
> may be unusable
> without the ability to efficiently discover resource instances.
>
> Actually, this is a problem even for NETCONF and RESTCONF, and there
> are list pagination
> drafts in progress to help fix it.
>
>
>
> Andy
>
>
> > module list-inst {
> >    prefix t;
> >    namespace "http://example.com";
> >
> >    list instance {
> >      key "name";
> >
> >      leaf "name" { type string; }
> >      leaf "value" { type uint32; }
> >    }
> > }
> >
> > How do you discover existence of let's say instance with name "hidden"
> > (i.e. "/list-inst:instance[name="hidden"]" 'instance-identifier')? You
> > can use a GET method but the overhead for non-trivial YANG models is
> > going to kill any reasonable real-world usage (IMHO). The overhead is
> > huge because the GET will send the whole datastore (meaning all
> > available instance data nodes).
> >
> > Also note that "/list-inst:instance" is not valid 'instance-identifier'
> > as it does not encode all list keys. The exactly same problem is with
> > using SIDs but less human readable. The only thing that changes is
> > instead of "/list-inst:instance" we use SID like 60,000.
> >
> > I find that the protocol should be extended with ability to enumerate
> > list instances and ability to get length of the instance list (and
> > leaf-lists).
> >
> > Just that I am curious, how is this done in RESTCONF?
> >
> >
> > Best regards,
> >
> > Vojtech Vilimek
> > CZ.NIC z.s.p.o.
> > Czech Republic
> >
> > _______________________________________________
> > core mailing list -- [e-mail]
> > To unsubscribe send an email to [e-mail]