Re: [core] CORECONF k parameter encoding

[Carsten Bormann <[e-mail]>]

On 2022-09-15, at 16:34, Koen Zandberg <[e-mail]> wrote:
> 
> One thing I'm running into, where I'm wondering if there has been previous discussion on this, is the encoding of YANG datatypes in the 'k' query parameter. There is a large variety of different encoding methods for the content, increasing the decoding complexity on the constrained device. Has it been consider of encoding all query params as a CBOR array and urlSafeBase64 as key value. The upside to this approach is that decoding the query parameter for a GET and DELETE requests closely matches the decoding of the payload with FETCH and iPATCH requests. Encoding the 'k' parameter also as CBOR array results in that for both a FETCH and a GET, the YANG list keys can in both cases be retrieved by iterating over a CBOR array.

Hi Koen,

The irregularity of the table at Section 4.1 of draft-ietf-core-comi [1] was exactly what I tried to bring up in [2].
I tried to find out reasons why it is so irregular and came up empty.

I like your suggestion for resolving this (except that I would maybe encode an RFC 8742 CBOR sequence, saving a byte or 4/3 characters in the base64url encoding).

It is late for this change, but possibly not too late.

Grüße, Carsten



[1]: https://tools.ietf.org/html/draft-ietf-core-comi-11#page-14
[2]: https://mailarchive.ietf.org/arch/msg/core/8ZLdtAGTlr7IMLT4ZT0YGFYVKmI/