Re: [Cbor] CDDL parsing questions

[Carsten Bormann <[e-mail]>]

On 2022-08-18, at 12:43, Toerless Eckert <[e-mail]> wrote:
> 
> Agreed. But it seems we do not agree how to call that level of receiver
> processing. I was trying to call it CDDL parsing. Before CBOR, we would
> have called it protocol parsing i guess.

Well, the terminology is all over the place here.

I prefer to reserve the term “parsing” to text-based protocols that are best handled with parser generators.

A protocol decoder often has two levels: the lexical syntax (length fields etc.), breaking up the bytes into, say, TLVs, and the TLV processor that creates a semantic representation.  Except that TLVs are rarely done in a particularly structured way, hence all the CVEs.

What you seem to be alluding to, is the ingestion of a CBOR data item in CBOR generic data model into the semantic categories that application wants.  CDDL can describe some, but not all of this process.

CDDL can be used to write complex grammars that require more look-ahead than one would like to have, e.g.

Message = Message1 / Message2

Message1 = [foo, bar, 1]

Message2 = [foo, bar, 2]

Don’t do that.

(A tool that flags excessive look-ahead requirements would be useful.
In this case, putting the discriminator up front is helpful:
Message1 = [1, foo, bar]
Message2 = [2, foo, bar]
Whether the look-ahead actually hurts depends on whether the processing of foo and bar depends on whether the message is a message1 or a message2.  In some cases, the form that nominally requires more look-ahead is easier to process, because there is no such dependency...
)

Grüße, Carsten