Re: [core] SenML FETCH/PATCH without globally unique names

As long as it is defined as optional, I don't see a problem.  I would worry
about possible issues if it was forbidden.

Jim

> -----Original Message-----
> From: core <core-bounces@ietf.org> On Behalf Of Ari Keränen
> Sent: Wednesday, March 6, 2019 7:03 AM
> To: core WG <core@ietf.org>
> Subject: [core] SenML FETCH/PATCH without globally unique names
> 
> CoRE WG,
> 
> One issue on SenML FETCH/PATCH that came up during the joint IETF / OMA
> meeting was that it would be useful to be able to use PATCH payload names
> without the "globally unique prefix".
> 
> SenML RFC has a requirement that all SenML names are globally unique [1].
> This enables the SenML Packs to be easily moved between systems without
> knowing the context they were originally retrieved at.
> 
> However, when SenML (i)PATCH is applied, the context is always known by
> the receiver since the receiver is hosting the resources that are being
> patched.
> 
> For example with LwM2M composite write using iPATCH, the content of the
> Patch Pack in JSON format could be as follows:
> 
> [{"bn":"/3303/0/", "n":"5700","v":22.4}, {"n":"5601","v":12.2},
> {"n":"5602","v":34.2}, {"bn":"/3303/1/", "n":"5700","v":18.2},
> {"n":"5601","v":9.2}, {"n":"5602","v":27.2}]
> 
> (Here you have a LwM2M client with two instances of a IPSO temperature
> sensor, Object ID 3303, supporting following resources 5700: Sensor Value,
> 5601: Min Measure Value, 5602: Max Measured Value).
> 
> If we keep the same requirement as SenML Pack names, both "bn" field
> values would have to be prefixed with the LwM2M endpoint ID, e.g.,
> "urn:dev:mac:0024befffe804ff1". This would consume a fair amount of bytes
> on wire.
> 
> Our suggestion is that we relax the requirement for the names in Patch
Packs
> so that the names are, for example, "evaluated in the context of the
target
> resource". That is, all base names could be automatically prefixed by the
> unique endpoint name and target resource path.
> 
> Comments welcome! This will be discussed in the CoRE interim in one hour.
> 
> 
> Cheers,
> Ari
> 
> [1] https://tools.ietf.org/html/rfc8428#section-4.2
> _______________________________________________
> core mailing list
> core@ietf.org
> https://www.ietf.org/mailman/listinfo/core