OpenPGP preimage gap, MPLS RSVP-SR fight, and protocol drafts on PQC and signaling
Standards work on 20 November centered on a second-preimage gap in the composite post-quantum OpenPGP signature draft and a contested MPLS adoption call. Parallel threads covered RPKI migration measurements, DNS and ACME signaling, AI use preferences, OAuth client fetch hardening, and unexplained Microsoft mail blocks.
OpenPGP composite signatures miss second-preimage resistance
The OpenPGP working group identified a second-preimage gap in the composite post-quantum signature draft (RFC 9980). Authors proposed raising minimum digest sizes and said they will rerun working-group last call. Implementers of composite signatures should treat current digest floors as insufficient until the revised text lands.
MPLS adoption call for RSVP-SR draws architecture objections
The MPLS working group closed an adoption call for draft-beeram-mpls-rsvp-sr-01 on 17 November after a large, heated thread. Opponents argued the draft conflicts with segment-routing architecture in RFC 8402 and RFC 8426 and reintroduces transit state. Adoption is not settled in the discussion record, and operators should treat RSVP-based SR as contested rather than consensus design.
SIDROPS asked to adopt PQC RPKI experiment report
The SIDROPS working group opened a call for adoption of draft-yoshikawa-sidrops-pqc-rpki-03, ending 2 December. The informational draft reports post-quantum signature experiments and RPKI migration measurements. It is measurement and migration data, not a protocol change, but it is the current venue for how RPKI might move off classical signatures.
Draft proposes client signals for ADoX transport preferences
An individual draft, draft-fernandez-dnsop-client-side-adox-signal, was posted to DNSOP and drew discussion of cache handling, integrity versus DNSSEC, and DELEG authentication. The draft would let clients tell resolvers which ADoX transports they prefer. Resolver operators should watch whether the signal is treated as a hint or as something that must be authenticated.
AI-CONTROL debates exemptions for user-directed generative use
The IETF AI-CONTROL working group debated whether the AI Use vocabulary should carry source preference signals for user-directed generative uses or exempt those uses in the style of copyright exceptions. The thread asked for concrete examples of non-consensual AI user usage that the vocabulary is meant to constrain. The outcome affects whether preference signals apply to ordinary user prompts or only to other classes of automated use.
ACME draft may shrink to dropping ES256 as mandatory
The ACME working group discussed shortening draft-ar-acme-pqc-tlsjws-01 to a minimal update of RFC 8555 that removes ES256 as mandatory to implement. The longer post-quantum JWS design is not what the thread settled on as the near-term change. Certificate authorities and ACME clients that still treat ES256 as required should expect that assumption to be challenged.
OAuth CIMD thread questions extra client-fetch hardening
The OAuth working group took up a comment from IETF 127 on whether the CIMD draft needs explicit client-fetch hardening beyond existing SSRF guidance. The thread did not record a decision. Implementers of client metadata fetch should not assume current SSRF text is the last word.
Operators report unexplained Microsoft S3150 blocks
Multiple operators on mailop reported Microsoft S3150 blocks with no useful detail and mitigation that did not clear the blocks. The reports describe an operational outage for senders rather than a published policy change. Mail operators hitting S3150 currently have little public diagnostic path.