Ladislav Lhotka <lho...@nic.cz> wrote:
> On Tue, 2018-11-06 at 07:36 +0100, Juergen Schoenwaelder wrote:
> > I agree that
> >
> > leaf datastore {
> > type ds:datastore-ref;
> > description "The identity of the datastore for which
> > the instance data is documented for config=true data nodes.
> > The leaf MAY be absent in which case the running dtastore or
> > if thats not writable, the candidate datastore is implied.
> >
> > For config=false data nodes always the operational
> > data store is implied.";
> > }
> >
> > is pretty confusing. It should be something like this:
> >
> > leaf datastore {
> > type ds:datastore-ref;
> > description "The identity of the datastore holding
> > the instance data. If the instance data is not associated
>
> Or rather the datastore that the instance data was extracted from.
I prefer "associated with". There are other uses cases than just
holding data "extracted from", e.g., data that is supposed to "be
inserted into" a datastore. "associated with" is less resrictive.
> After that,
> the data exists on its own and the originating datastore may later be holding
> something else.
>
> > with a datastore, then this leaf MUST be absent.";
>
> RFC 2119 language would make sense if there is anything that could break if
> that
> MUST isn't observed. But we even didn't know what the data is going to be used
> for.
>
> I would treat the "datastore" item as a purely optional information
I agree.
/martin
> that, if
> present, was provided for some reason. If it is false, what can we do?
>
> > }
> >
> > I am against merging data from different datastores together, which
> > the last sentence of the original text seems to imply.
>
> Both config true and config false data may come from <operational>, so it
> doesn't necessarily mean any mixing of datastores. But then again, I think
> that
> the datastore information isn't in most cases that interesting.
>
> Lada
>
> >
> > /js
> >
> > On Tue, Nov 06, 2018 at 11:51:26AM +0700, Ladislav Lhotka wrote:
> > > Joe Clarke <jcla...@cisco.com> writes:
> > > > ===
> > > >
> > > > Section 6
> > > >
> > > > With your datastore leaf, if I pull this off of a running YANG server,
> > > > serialize it and share it with my customer, why wouldn't I have the
> > > > actual datastore from which I retrieved it? What I'm saying is that
> > > > this element may be missing, but if it is, I don't think you can assume
> > > > the source datastore for config=true nodes.
> > > >
> > >
> > > The description of the "datastore" leaf doesn't make much sense to
> > > me. It says that for configuration data the default is "running" or
> > > "candidate" if "running" isn't writable. Why should it matter whether
> > > "running" is writable? It looks like it is assumed that the config data
> > > will
> > > eventually be fed into the indicated datastore, but I don't see any
> > > reason for such an assumption.
> > >
> > > I can see that "datastore" can be occasionally useful as auxiliary
> > > metadata but, in my view, this document addresses also instance data
> > > that is not necessarily bound to any datastore.
> > >
> > > Lada
> > >
> > > --
> > > Ladislav Lhotka
> > > Head, CZ.NIC Labs
> > > PGP Key ID: 0xB8F92B08A9F76C67
> > >
> > > _______________________________________________
> > > netmod mailing list
> > > netmod@ietf.org
> > > https://www.ietf.org/mailman/listinfo/netmod
> --
> Ladislav Lhotka
> Head, CZ.NIC Labs
> PGP Key ID: 0xB8F92B08A9F76C67
>
> _______________________________________________
> netmod mailing list
> netmod@ietf.org
> https://www.ietf.org/mailman/listinfo/netmod
>
_______________________________________________
netmod mailing list
netmod@ietf.org
https://www.ietf.org/mailman/listinfo/netmod
