[Multiformats] IETF 116 Dispatch Multiformats: Responses to Harald Alvestrand

Manu Sporny <msporny@digitalbazaar.com> Mon, 10 April 2023 15:31 UTC

MIME-Version: 1.0
From: Manu Sporny <msporny@digitalbazaar.com>
Date: Mon, 10 Apr 2023 11:30:58 -0400
Message-ID: <CAMBN2CTJHSVsS=jxLtN2xS_pgs6GJ4VhTZM=UMCijxiYvJWyZQ@mail.gmail.com>
To: multiformats@ietf.org
Cc: Harald Alvestrand <harald@alvestrand.no>
Content-Type: text/plain; charset="UTF-8"
Archived-At: <https://mailarchive.ietf.org/arch/msg/multiformats/8xGsAAdXspPIebVzJgJMwG_5wOE>
Subject: [Multiformats] IETF 116 Dispatch Multiformats: Responses to Harald Alvestrand
Precedence: list

Harald Alvestrand said:
> Multiformats are just a little part of a system, so it's hard to evaluate

Yes, this is true. Multiformats are used in application data formats
and so analyzing them without understanding how they fit into the
bigger picture can be challenging. An analogy for similar IETF work
would be RFC6920 "Naming Things with Hashes" or
draft-ietf-httpbis-digest-headers "Content-Digest", which are used in
application protocols.

That said, Multiformats are being used in the IPFS community (in their
data formats and protocols), as well as in W3C specifications (in data
models and protocols). Some examples of usage are here:

https://www.w3.org/TR/vc-data-integrity/#multikey
https://docs.ipfs.tech/concepts/content-addressing/#what-is-a-cid

> Registration registers keys in a very small space

As mentioned during Dispatch, this was a miscommunication around the
registration space for Multiformats. All Multiformats use a
variable-length integer, encoded as a varint, making the universal
namespace for types theoretically unlimited (in practice, for security
reasons, implementations limit the varint length to 9 bytes; 63 bits
of value space).

This means that lower numbers in the table can be expressed as a
single byte, whereas larger numbers in the table might need more than
one byte to be represented. The total size of the Multiformats
namespace is 9,223,372,036,854,775,807 multiformat types.

So, while registration procedures are needed, they don't need to be
overbearing since the value space is probably more than we'll ever
need (at least, for the next century or so).

> All usages cited are W3C-related, one option is to do this at W3C
> It's not clear if IETF work needs Multiformats

We've had a few discussions with Roberto and Lucas in the HTTP WG
around the use of Multibase and Multihash in the Content-Digest header
(draft-ietf-httpbis-digest-headers). Our company also uses
Multiformats to refer to public key information in the HTTP Message
Signatures specification. There is some overlap with IETF work, though
it's minor at this point in time.

We had considered going W3C with these specifications, but "spinning
up a light weight WG" is not something that W3C does.

We did also consider just creating a registry at W3C through the W3C
Verifiable Credentials WG, but it was suggested that since this is a
much more "low level" specification, that it belongs more at IETF than
it does W3C.

All this to say, we could create and maintain a registry for this
stuff at W3C, but many of the community participants felt that the
work aligned more with an IETF RFC than it would a W3C REC or
Registry.

I hope that answers your concerns, Harald, and if not, happy to try
and provide more insight if something was left unanswered.

-- manu

-- 
Manu Sporny - https://www.linkedin.com/in/manusporny/
Founder/CEO - Digital Bazaar, Inc.
News: Digital Bazaar Announces New Case Studies (2021)
https://www.digitalbazaar.com/

[Multiformats] IETF 116 Dispatch Multiformats: Re… Manu Sporny