Re: [netmod] CBOR tags for Common YANG Data Types (RFC6991/bis)

[Andy Bierman <[e-mail]>]

On Sat, Jul 30, 2022 at 10:05 PM Carsten Bormann <[e-mail]> wrote:

> On 31. Jul 2022, at 02:12, Andy Bierman <[e-mail]> wrote:
> >
> >     leaf foo {
> >         type inet:ipv6-address;
> >         ext:cbor-type cbor:bin-ipv6-address;
> >     }
>
> This looks like the right thing to do.
> But it touches many moving parts, and I’m wondering whether we cannot do
> something with a more localized footprint.
>
> This is all about the representation; the data model doesn’t actually
> change (*).
> We don’t need to change the model to go from XML to JSON, why should we
> need to do that here?
>
> Maybe this could be done via a concept that looks more like a SID file.
> It would need to be visible in a yang-library-like model.
>
>
Agreed.
Both sender and receiver need to use the same "replacement encodings".
E.g, the SID file entry indicates certain objects that should use
date-and-time
are flagged to use Tag 1 instead.

If a YANG model is used to identify the server-specific replacements,
then the client code cannot be hard-wired at compile-time.
It has to be adjusted at run-time for each server.
I know it has been a priority for some WG members that a client should
be able to code to the SID files without any additional (run-time) data.


Grüße, Carsten
>

Andy


>
>
> (*) Well, maybe it does: E.g., when using Tag 1 for a date/time, there is
> no way to represent a numerical offset.
> That is generally not needed, so the conversion would work in the majority
> of cases.
> If a numerical offset needs to be expressed, the text-based form could be
> used.
>
>