Re: [fdt] [Webtransport] Standards for protocol headers?

[Carsten Bormann <[e-mail]>]

On 2020-06-16, at 13:46, Marc Petit-Huguenin <[e-mail]> wrote:
> 
> The next logical step in the evolution (or is it devolution?) of XML then JSON is SExp.

Hi Marc,

This is actually a quite profound observation.

We used S-Expressions in the work that led up to RFC 3259, and everything went downhill since then.

The important part of an S-Expression design is, of course, what kind of data goes in there; the nice thing about S-Expressions is that, to a certain extent, you get to design that in a bespoke way for each case, but that also is its downfall as it cannot be standardized.

JSON (or, preferably, CBOR) actually makes nice S-Expressions if you just don’t use those maps (“JSON Objects”); see https://www.ietf.org/id/draft-bormann-cbor-cddl-freezer-04.html#name-alternative-representations for a current example (yes, that is the entirety of CDDL in that figure).  Not being able to separate strings and symbols does hurt, but can be worked around.

Grüße, Carsten