Re: [dispatch] IETF 116 - do you have something for DISPATCH?

[Christopher Allen <[e-mail]>]

On Thu, Mar 2, 2023 at 9:02 AM Dale R. Worley <[e-mail]> wrote:

> Contacting the chair(s) of the CBOR WG(s) would be a start.  You might
> also search the IETF mailing list archive for "dCBOR" and "deterministic
> CBOR" to see what mailing lists have discussed it.  Or joining any
> relevant WGs and posting notices to their mailing lists.
>

Thanks. We have attempted to contact various chairs, but given the
challenges of mail volume, spam, and centralization of mail providers, I
worry that the lack of response to date from them, that they are not even
seeing the emails. For instance, I've found with IETF lists that I can't
use gmail the way I normally do, and have to send differently.

> Finally, we would like advice from Dispatch on the best way to register
> our
> > cryptographic-focused CBOR tags used in Envelope with IANA, namely
> whether
> > we should do so now, or whether we should wait for later in the IETF
> > process.
>
> I've followed the CBOR work some, and in general there's no shortage of
> tags.  So registering it early is likely to be OK.  I see from
> https://www.iana.org/assignments/cbor-tags/cbor-tags.xhtml that tags
> <32768 are "Specification Required" and tags >=32768 are "First come
> first served".  That suggests that if you expect to use a tag, and the
> tag already has a clear, frozen definition, you might as well register
> it now with its specification.  This works especially well if the tag
> may have uses in other applications.
>

We are hoping to at least get tags in the smaller 200 range, as the
requirements for tags (in theory) suggest that a published spec is
sufficient, and thus our I-D should qualify (there are certainly others in
that range that don't even have that!). We have listed our requests for
them in our I-D under IANA considerations
https://www.ietf.org/id/draft-mcnally-envelope-00.html#name-iana-considerations


However, in particular with the core envelope tag, having it be even
smaller could be quite advantageous, as some of our developers are using
very constrained silicon where every byte counts. As the design of
envelopes is "envelopes of envelopes," there can be quite a few of these
tags. This would require support from a working group (or another standards
org, which may also be possible).

OTOH, if further work on Gordian Envelope is likely to change the
> details of the specifications of your tag, you aren't violating the
> rules if you register it FCFS; few will care if you update the
> definition or abandon it entirely.  (Does the length of the tag matter
> in your usage?)
>
> ("Gordian Envelope" sounds similar to letterlocking, which has been in
> the blogosphere recently!)
>

Yes! In concept, it is. We have a a short video introducing our concepts at
https://www.youtube.com/watch?v=OcnpYqHn8NQ, and you'll see some
paper-world examples. I've been seeking a CC-BY photo of a letter locking
envelope to add to our imagery ;-)

-- Christopher Allen