cbor

Summary

The CBOR working group met at IETF 125 to primarily discuss the progress of draft-ietf-cbor-serialization. The session focused on the recent changes in version -04, the inclusion of examples (formerly referred to as test vectors), and the handling of NaN payloads. The group also reached a decision to "unpark" other working group documents, specifically draft-ietf-cbor-edn-literals, due to external normative dependencies.

Key Discussion Points

Serialization Draft Status

Laurence Lundblade presented the Serialization draft status and discussion. Key updates in draft-ietf-cbor-serialization-04 include:

Renaming "ordinary serialization" to PreferredPlus.
Removal of Appendix C (deterministic serializations for common tags).
Renaming control operators (e.g., using .det was avoided as it was taken).
Expansion of the introduction and wording improvements based on reviews by Rohan Mahy.

Ongoing issues include a PR from Carsten Bormann regarding Section 1 and a proposal to move the NaN appendix to a separate document. Carsten Bormann noted that Rohan Mahy had sent additional change requests for the Section 1 PR.

Examples and Test Vectors

Laurence Lundblade proposed including a curated set of approximately 13 examples in draft-ietf-cbor-serialization to illustrate General, PreferredPlus, and Deterministic serializations.

Terminology: Dave Thaler and Carsten Bormann argued that these should be called "examples" rather than "test vectors" to avoid confusion with exhaustive test suites. Examples are seen as vital for visual learners and for clarifying the prose.
Technical Content: The examples cover integers, maps with multiple possible orders, arrays, and NaN payloads.
Decoder Behavior: Discussion ensued regarding how non-checking decoders should handle serializations not matching their preferred mode (e.g., a PreferredPlus decoder encountering a NaN payload). Laurence Lundblade noted that a non-checking decoder's response to non-compliant input is flexible (it could error out or decode it) as long as it correctly decodes its required serialization profile.
Location: Hank Birkholz suggested that while a few examples belong in the draft, larger sets should reside in a GitHub repository. Barry Leiba suggested the possibility of a separate informational document for test vectors, though Laurence Lundblade and Carsten Bormann emphasized that examples are necessary within the main draft to ensure the framing and technical rationale are correct.

Document Roadmap and "Unparking"

The chairs and the group discussed the status of documents currently on hold.

EDN Literals: Rohan Mahy and Andy Newton pointed out that SD-JWT/SD-COT has a normative dependency on draft-ietf-cbor-edn-literals.
Packed CBOR: There is a desire to resume work on draft-ietf-cbor-packed once the serialization work reaches a stable point.

Decisions and Action Items

Decision: The working group will "unpark" draft-ietf-cbor-edn-literals to move it toward completion alongside the serialization work.
Action Item: Rohan Mahy to provide additional security considerations for the serialization draft by the end of the week.
Action Item: Laurence Lundblade to produce draft-ietf-cbor-serialization-05 within approximately two weeks, incorporating the curated examples and addressing Carsten Bormann's PR.

Next Steps

The Chairs (Barry Leiba, Christian Amsüss, and Francesca Palombini) will formalize the updated work plan on the mailing list.
Work will resume on draft-ietf-cbor-edn-literals followed by draft-ietf-cbor-packed.

Automatic IETF Minutes