Re: [core] Simplify draft-ietf-core-comi-12.txt

On Sat, Mar 18, 2023 at 2:23 PM Carsten Bormann <cabo@tzi.org> wrote:

> In the interim last Wednesday, we were seriously wondering why we need
> both GET/PUT/POST and FETCH/iPATCH for COMI.  The consensus of the room was
> that the easiest way to solve the problems with generating the URIs for
> GET/PUT/POST was to move over to FETCH/iPATCH completely.
>
>

I support the move to FETCH/iPATCH.
(We called it GET/SET back in the day ;-).

The CORE WG has reinvented SNMP and fixed almost everything wrong with it.

-  Varbinds can for any object, not just a leaf
-  Data Model is YANG (replacing legacy SMIv2)
-  Data retrieval is super-efficient instead of extremely inefficient
-  Data representation has no duplication instead of massive key replication
-  Configuration editing based on subtrees (not leafs) is now possible,
-  Capabilities and Error Reporting are much improved and extensible

IMO the protocol should be as lean as possible.
Take out (or make optional) everything that is not needed.

Andy

I have created a first draft of this “simplified” version of core-comi.
> No base64 encoding any more…
> All CoAP method/content-format combinations that would have required
> base64-encoding instance identifiers on a URI have been removed.
>
> GET/PUT/POST/DELETE are still in there for full datastore access.
> I’m not sure about the use case for this, but since it doesn’t need
> instance identifiers in a URI, I left it in.  Please advise.
>
> I also simplified the POST request for RPC/Action to carry
> application/yang-instances+cbor as a payload.
> This was a single "application/yang-data+cbor; id=sid", but this is now no
> longer used by CORECONF (*) (and it doesn't carry an instance identifier).
>
> event-streams are still fetched using GET as all the information that is
> needed is in
> the event stream URI, but I don’t know a good way to do ?f= on event
> stream resources.  Should this be moved to FETCH as well?
>
> (*) I don’t know how to do errors in the way COMI has done them before.
> These don’t cleanly map to application/yang-instances+cbor, as there is no
> data node involved.  (By now, we have RFC 9290…  Does this help here?)
>
> Please view the draft draft at:
>
> https://core-wg.github.io/comi/simplify/draft-ietf-core-comi.html
>
> A diff from main (~ -12) can be seen at:
>
>
> https://author-tools.ietf.org/iddiff?url1=https://core-wg.github.io/comi/draft-ietf-core-comi.txt&url2=https://core-wg.github.io/comi/simplify/draft-ietf-core-comi.txt
>
> Enjoy!
>
> Grüße, Carsten
>
> _______________________________________________
> core mailing list
> core@ietf.org
> https://www.ietf.org/mailman/listinfo/core
>