Re: [netconf] [Technical Errata Reported] RFC8040 (6342)

[Andy Bierman <[e-mail]>]

On Mon, Nov 23, 2020 at 9:27 AM Kent Watsen <[e-mail]> wrote:

> +1 on rejecting the errata.  PUT replaces the resource, PATCH modifies it.
>
> Any node not present in a PUT representation is assumed to be deleted, but
> it is not allowed to delete keys, hence why keys MUST always to present in
> a PUT.
>
>
> Although we have several customers that would like to rename list entries,
> and sometimes think RESTCONF allows it.  I do not think any YANG-based
> protocol
> allows list keys to be modified.  You have to delete and re-create the
> list entry.
> I think this is something for YANG-next to consider, so it is done
> consistently across
> all protocols.
>
>
> YANG-next or RESTCONF-next?
>
> I think that it is unique to RESTCONF since only RC (not considering COAP)
> targets specific resources.
>
>

I think any YANG list with keys can have this problem.
IMO it is better to define the common semantics in YANG and let the
protocols
decide how to encode them on the wire.

The workaround (of course) is to pick an arbitrary integer or label as the
key
so there is no reason to change it.


That said, the feature might be something that could be added via an I-D,
> as opposed to an 8440-bis.
>
>
Not to restart this thread, but I prefer yang-next vs. lots of add-on RFCs.
Yes it is harder for standards writers to produce, but easier to use in the
long run
for the vendors and operators.



> K.
>
>
Andy