ATP Hybrid Meeting @IETF-126

Thursday 23 July 2026
Room: Grand Klimt Hall 1
16:30 - 18:30 Local
Log into the IETF datatracker to access:

• MeetEcho
• Notes
• Zulip

ATP Meeting Agenda

16:30 Status and Agenda Bash - Chairs and ADs (10 min)

Shuping (co-Chair) suggested to update the milestones to reflect the
discussions in the mailing list.
The Repo and Sync drafts will be worked together, and the authors
suggested to merge the two into one.
The account identifier is suggested to be worked on first and then the
URI draft.

Justin suggested that the milestones can be left as they are now. The
work will be driven by the discussions in the working group.

Andy (AD) suggested to add dates to the milestones later.

Shuping asked that the new drafts submitted to the working group need to
use the -atp- tag in order to be automatically added to the datatracker.

16:40 Status of Repository and Synchronization documents (20 min)

Presenter: Bryan Newbold (On site)
draft-holmgren-at-repository
draft-holmgren-at-synchronization

[Ted Hardie] - recommends privacy considerations belongs in the
architecure doc, and not to be bundled with security (based on lessons
learnt from other IETF groups)
[Ted Hardie] - recommends not changing HTTP API endpoints until
there's a non backward compatible change
[Phillip Hallam-Baker] - recommends moving to well-known endpoints

17:00 Possible Changes for IETF Version (20 min)

Presenter: Daniel Holmgren (On site)
Discussion

[Eric Rescorla] - recommends removing secp256k1, not bothering with
ed25519, move towards MLDSA
[Ben Go] - questions whether to have these signature algorithms
specified within the doc
[David Schinazi] - recommends keeping a signature algorithm registry
within doc, keeping it to a short list, mandating implementations
upgrade to changes

[Tom Hert] - questions whether time based cursors continue to work
across providers as the ecosystem grows

17:20 The AT URI Syntax Challenge (20 min)

Presenter: Bryan Newbold (On site)
The AT-URI Syntax Mess

[Ted Hardie] - recommends at:did... , strongly recommends against
other options
[Ted Hardie] - if no longer calling these URIs, this will need to be
well signposted
[David Schinazi] & [Justin Richer] - recommend against pushing for
URI changes
[Markus Sabadello] - supports tripple slash option, warns that DID's
can contain paths
[Phillip Hallam-Baker] - recommends at:did..., calls out the URI slash
formats was for relative linking
[Jan Linblad] - called out that percent encoding may need to be
supported anyway

17:40 ATMoQ - an implementation of an atproto relay over the MoQT (15 min)

Presenter: Eli Mallon (Remote/On site)
ATOM: AT Protocol Over MoQ Transport

[Emelia Smith] - asked about wider usage of the conformance testing

17:55 Block Ordering (15 min)

Presenter: Phil Schleihauf (Online)
Discussion

[Brian Newbold] - concern that relying on decimal points or scientific
notation string parsing to infer types makes the protocol fragile, as
standard JSON parsers may normalize 1.0 to 1 before ATP can parse
it.
[Daniel Holmgren] - suggested placing the canonicalization logic
entirely in the CBOR encoder to prevent requiring custom JSON parsers.

18:10 Floats question (15 min)

Presenter: Bumblefudge (On site)
The tag-42 profile of CBOR

[Bryan Newbold] - concerned with relying on JSON data assumptions that
aren't specified in JSON
[Bryan Newbold] - recommends relying on the CBOR representation rather
than parsing the JSON to check hash / signatures
[Daniel Holmgren] - concerned with requiring custom JSON parsing, but
would prefer if custom requirements stays in the CBOR encoder

Flextime & AOB (5 min)