Re: [Cbor] Review of draft-ietf-cbor-packed-02

[Carsten Bormann <[e-mail]>]

Hi Russ,

thank you for this review.

> Document: draft-ietf-cbor-packed-02
> Reviewer: Russ Housley
> Review Date: 2021-03-15
> 
> Major Concern:
> 
> RFC 7049 says that CBOR was designed so that "encoded data should be
> self-describing so that a generic decoder can be written."  I di not
> expect Packed CBOR to deviate from this design goal.  However, the
> feature that allows an application environment, such as a media type,
> to define tables does exactly that.  Decode will fail if these table
> values are not known to the decoder.

Well, decode won’t fail, but unpacking will leave holes.
That is a consequence of using external dictionaries that is hard to avoid.
Would you prefer to rule out external dictionaries?  
They sure were useful in places like RFC 3485 etc.

> Minor Concern:
> 
> In Section 2.1, the text says: "In the abstract, ...".  I am unsure what
> this is trying to tell the implementer.  I think it means: "regardless
> of the data structure used by an implementer".  I think these words can
> be removed.

I think this was supposed to mean:

>> Each of the tables is indexed by an unsigned integer (starting
>> from 0), which may be computed from information in tags and their
>> content as well as from CBOR simple values.


> Nits:

> 
> Section 2: Begins with: "Packed CBOR is defined in two parts:
> Referencing packing tables (this section) and setting up packing
> tables (Section 3)."  I suggest this text be place at the send of
> the Introduction.

Changed this to:

>> Packed CBOR is defined in two parts: Referencing packing tables
>> ({{sec-packed-cbor}} and setting up packing tables
>> ({{sec-table-setup}}).
(At the end of 1. proper; Terminology section interspersed…)
>> # Packed CBOR
>> 
>> This section describes the packing tables, their structure and how
>> they are referenced.

Both changes are now in https://github.com/cbor-wg/cbor-packed/pull/2

Thank you!

Grüße, Carsten