Re: [core] YANG to CBOR mapping version 1

[peter van der Stok <[e-mail]>]

Alexander Pelov schreef op 2016-07-14 10:39:
> Dear Peter,
> 
> I agree with you - if a server expects TWO keys, but receives ONE this
> is an error. What I don’t understand is how marking the CBOR content
> changes anything to this. (if the client insists that a leaf is a key
> and the server disagrees - then you have error. if the server requires
> a key and the client does not provide it - stil an error).
> 
> From implementation point of view, the server will never recompile the
> YANG module. YANG is used compile-time (when you generate your program
> on your laptop) and as a contract between the client and the server.
> The same goes for constrained clients.
> 
> In most cases, the server and the client will actually not need to
> have the YANG module scheme at all - only a couple of SIDs, associated
> meta-data (in binary - such as the type) and the associated functions
> (which you need anyways). As SIDs never change, you don’t need
> anything else on the device. It could be even simpler - without the
> meta-data - but then the associated functions need to do the
> validation (e.g. Arduino implementations).
> 
> Alexander

Hi Alexander,

absolutely agree with you in the above.
The need to identify keys and to distinguish different instances is 
needed when you want to "replace" (same key values) or "add" (different 
key values) individual instances to a list.

I try to argue that you want that express this in the payload because 
you cannot count on server and client to have the same view on the key 
set.

Peter