Re: [core] YANG to CBOR , identification of objects

[Carsten Bormann <[e-mail]>]

Michel Veillette wrote:
> do you have a strong opinion about the approach proposed by Peter?

Actually I'm intrigued as to what that approach will be; until we know
that, I'd be a bit weary about moving things around.

So far, I have been quite happy with the parent delta approach, but I'm
not sure we have really succeeded in explaining it succinctly and
unambiguously.
For instance, what happens if a data structure is buried further into
another one that itself doesn't have a SID?

I'm not a big fan of specification-by-algorithm, but here it probably
*would* help to phrase this as an algorithm (preferably functional).
Cf. RFC 7396: The pseudo-code on page 4 really helps understanding the
obtuse language around it.

Given that the CBOR data model maps right into data structures that can
be addressed by pseudo code, specifying the "decoder" (as is usual for
compression algorithms) should be lean and precise.

Grüße, Carsten