[Cbor] Some comments on cbor-cddl-05 and cbor-7049bis-02

[John Mattsson <[e-mail]>]

Hi,

Hi, took some time to read these two documents in detail during my last trip. Both documents are very well written and quite easy to read and understand. Some comments:


Comments on cbor-cddl-05
---------------------------------

- I think cbor-cddl-05 should formally update RFC7049 or RFC7049bis as it extends the diagnostic notation.

- Section 2.2.1 seem to define value notation for integers, floats, and text strings, but not byte strings (which in my use cases are often used very much). Maybe the cddl tool already implements something that is just missings from the documents, otherwise all or some of the diagnostic notation for byte string would fit. e.g. h'0102030405'

- If I understand correctly, Section 3.8.5. uses the $ notation which is not defined until Section 3.9.

- That ".ctrl" in Table 1 means all CDDL control operators should be explained.

- Section 2.1.2 states that "comma is actually optional (not just in the final entry), but it is considered good style to set it."

Reading the doc it feels like the good style is to set all commas in definitions with a line break after each entry, but to leave out the last comma in definitions that do not use line breaks after each entry

orderedpair = [
   a: int,
   b: int,
]

orderedpair = [ a: int, b: int ]

I think the sentence should be updated to reflect this.

- More guidelines or at least links on how to include CDDL in RFCs would be helpful, i.e.

   <artwork type="CDDL"> in xml
   
   ~~~~~~~~~~~ CDDL in kramdown

- SenML is now published as RFC 8428

- The document refers to the cddl tool, it could also refer to tools for diagnostic notation in the extended diagnostic notation section, e.g. cbor.me


Comments on cbor-7049bis-02
---------------------------------

- 7049bis should obsolete RFC7049, I think.

- The terminology section should refer to RFC 8174

- JSON should refer to RFC 8259 instead of RFC 7159

- RFC 5226 should be replaced by RFC 8126

- ANS.1 should refer to the 2015 version
  https://www.itu.int/itu-t/recommendations/rec.aspx?rec=x.690

- ECMA/262 should refer to the 9th edition, 2018

- [TIME_T] should refer to the 2017 edition

- Could refer to tools for diagnostic notation in the extended diagnostic notation section, e.g. cbor.me


Comments on cbor-7049bis-02 or cbor-cddl-05
---------------------------------

I think it's a large benefit if the CDDL and the CBOR diagnostic notation is as similar as possible. The diagnostic notation should allow comma after the final entry. The specification is not clear here, but at least cbor.me dislikes all the following

[ 1, ]
{ 1, }
<< 1, >>

Cheers,
John