Hedge notes from IETF 119 STIR WG

Secure Telephony Identity Revisited

Chairs

1) Administrivia
- Agenda Bashing
- Minute Taker
- Jabber Scribe
- Bluesheets - Meetecho tool

- draft-wendt-stir-certificate-transparency-00 
    - Chris Wendt

3) STIR+MLS
- draft-peterson-stir-mls-01
- Jon Peterson and Richard Barnes

4) Any Other Business (if time permits)

07 is now out! (as of this morning) building in stapling.
Examples included.
Request that implementers review the new draft to evaluate the
impact of the size of the staple.
Question for attendees/reviewers: should there be a decoding of the
PEM in the example
Question for attendees/reviewers: should there be something
specifying the algorithms in the PASSporT, or is that built in to
OCSP?

No new version, since it's being called for adoption already.
Eric Rescorla has sent feedback on the mailing list
- Questions about x5u/x5c (and typo of x5y)
- Suggestion for highlight of normative change (move from x5u to
  x5c)
- Should certs be listed in both x5u AND x5c?
Alec Fenichel
- nothing precluding users from doing both if the URL is in both.
  Jon Peterson doesn't see an issue
- Example in the text (for OCSP) - mismatch in certificate
  extension (example ends .cer, frequently .pem is the extension
  for PEM format)
  - Jon Peterson to look at RFC 5280 to make sure this draft
    doesn't conflict!
  - Russ Housley to find a reference about specifying a chain of
    certs in DER.
Summary of direction forwards:
- Stick with PEM
- Change examples to reflect
- Keep x5c as written up but also add note to keep the x5u field
  in those shortlived
No objections for adoption of the short-lived draft.

Chris Wendt presenting
- One key decision looking for input on: Current certificate
  transparency document provides three patterns, which we might be
  able to simplify - do we need any mechanisms other than
  pre-certificates? (See RFC 9162)
Orie Steele (in chat) offered route to using just the "SCT" part
without the rest of RFC 9162, which is rarely implemented.
- "SCITT helped COSE lift the proof system RFC9162, its currently
  being specified for COSE in
  https://datatracker.ietf.org/doc/draft-ietf-cose-merkle-tree-proofs/"
Jon Peterson:
- This only works if everyone signs up otherwise a rogue CA
  doesn't list their certs (or their real certs) and all verifiers
  choose to trust.
- Recommends use of the 'observatory' approach, based on the way
  the old SSL Observatory used to work.
  - Observes and detects changes to existing certificates in use
    when that's not expected to flag suspicious new actors.
- Chris: could be applied primarily to delegates, where we don't
  have as many established processes today.
  - Jon: Ownership of individual numbers is complex, makes this
    harder.
Eric Rescorla: needs a threat analysis before this can be adopted
Jon Peterson: needs more work on the list, not ready for adoption.
Conclusion: not being adopted at this time, will continue
discussion.

Jon Peterson and Richard Barnes
Interaction with the MIMI WG (Jon Peterson also talking there.
More coordination and integration to be done.
draft-peterson-mimi-idprover is a strawman draft covering the
discovery problem in MIMI, reuses some ideas from ACME integration.
Conclusion: Too early for adoption, more work to be done in
MIMI.

Time does not permit!
What's the future of the WG? Are we on the long tail for closing
down?
- Call for discussion on the mailing list.