CBOR

Summary

The CBOR Working Group met during an interim session to discuss the status of draft-ietf-cbor-edn-literals following its Working Group Last Call (WGLC), and to review pending technical issues for draft-ietf-cbor-serialization.

The chairs outlined the path forward for the EDN literals draft, establishing a structured process for capturing last-call consensus. The group also reviewed four distinct technical updates to the serialization draft, including CDDL control operators, introductory terminology, sample float-encoding implementations, and pending security considerations.

Key Discussion Points

1. WGLC Status and Next Steps for draft-ietf-cbor-edn-literals

Feedback Process: Paul Hoffman reviewed the process for addressing WGLC comments. The chairs compiled a list of proposed changes and removals. Paul Hoffman acknowledged omitting a comment from Вадим Гончаров and notes that some feedback from Mikolai Gütschow required clarification.
Paul Hoffman will distribute a revised list of issues to the mailing list and restart a one-week timer for community feedback.
Consensus Rules: Carsten Bormann and Paul Hoffman emphasized that silence on the list indicates support for the draft as currently written, rather than support for proposed changes. WG members must explicitly "+1" or endorse proposed modifications on the list to establish rough consensus. Discussion on the "how" of changes should be avoided in favor of indicating "what" is critically important.
Editorial vs. Technical Changes: Carsten Bormann raised concerns regarding pending editorial work (e.g., ABNF restructuring, shifting text between sections, and simplifying terms like "application-oriented extension"). Paul Hoffman clarified that no distinction will be made between editorial and technical changes for this final push, as past editorial changes have carried technical weight. Any critical editorial adjustments must be brought to the list immediately.
Resolution Path: If there is rough consensus on an issue, Carsten Bormann will update the draft. If no rough consensus emerges for any listed item, the current version will be finalized and advanced to the IESG.

2. Status of draft-ietf-cbor-serialization

Laurence Lundblade presented open issues for the serialization draft using the slide deck Serialization Draft Status.

A. CDDL Control Operators

Problem: Previous iterations of the draft used .pref-p and .dtrm to indicate preferred or deterministic serialization. This was structurally invalid in CDDL because control operators require a right-hand side.
Proposed Solution: Laurence Lundblade proposed a .ser-operator control operator where the specific serialization constraint is defined on the right-hand side. This allows the constraint to apply to an entire CDDL document or arbitrary segments, which works independently of draft-ietf-cbor-cddl-modules.
Alternatives: Christian Amsüss proposed an alternative on GitHub where the right-hand side of .cbor or .cbor-seq is set to any instead of a byte string. Carsten Bormann expressed concern that .serial may not align with CDDL design conventions and suggested utilizing CDDL directives instead.
Discussion: Rohan Mahy suggested that if CDDL integration delays the serialization draft, the control operators could be extracted and addressed in a future, standalone CDDL extension. Laurence Lundblade agreed to take the discussion to the mailing list.

B. Section 1 Terminology and Intro Changes

Carsten Bormann submitted a Pull Request (PR) proposing terminology changes (specifically introducing "serialization constraint" instead of "serialization") and adding detailed introductory text.
Laurence Lundblade created an alternative, simplified PR that adopts some editorial changes but maintains a high-level, concise introduction. Laurence Lundblade noted that a formal terminology change requires broader working group consensus.
Paul Hoffman suggested that Laurence Lundblade post both text options to the mailing list for the working group to evaluate.

C. Example Source Code for Encoding Floats

Problem: While RFC 8949 contains sample code for decoding, it lacks sample code for encoding floats (converting double-precision to single-precision, and single-precision to half-precision).
Proposed Solution: Laurence Lundblade authored a PR containing C sample code designed for preferred and deterministic serialization.
- The implementation restricts conversion to trivial (quiet) NaNs, erroring out on non-trivial NaNs (differing from Carsten Bormann’s CDE draft implementation).
- It provides two distinct functions (double-to-single and single-to-half) to handle all necessary conversion pathways.
Carsten Bormann expressed appreciation for the two-step conversion model. Laurence Lundblade noted that while his proposed code is longer than Carsten Bormann's terse CDE version, it is much easier to understand and port to other programming languages while compiling to an equivalent object code size (~150 bytes).

D. Security Considerations

Security considerations text is pending. Rohan Mahy committed to drafting this section shortly.

3. Future Planning

Paul Hoffman noted that the CBOR session at the upcoming face-to-face meeting in Vienna is scheduled for Thursday from 11:30 to 12:30 (Europe time), with no major agenda conflicts.

Decisions and Action Items

Action Item: Paul Hoffman to distribute the updated list of draft-ietf-cbor-edn-literals last-call issues and restart the one-week feedback timer.
Action Item: Mikolai Gütschow to clarify on the mailing list which of his comments should be formally added to the EDN literals issues list.
Action Item: Carsten Bormann to submit any critical editorial or terminology modifications for draft-ietf-cbor-edn-literals to the list immediately.
Action Item: Laurence Lundblade to initiate a mailing list thread regarding CDDL control operators for draft-ietf-cbor-serialization.
Action Item: Laurence Lundblade to post the two alternative draft texts for Section 1 of draft-ietf-cbor-serialization to the list for WG review.
Action Item: Rohan Mahy to author the security considerations section for draft-ietf-cbor-serialization.

Next Steps

Resolve the remaining issues on draft-ietf-cbor-edn-literals within the next week to prepare the document for submission to the IESG.
Drive consensus on the outstanding issues for draft-ietf-cbor-serialization to position it for a Working Group Last Call in the coming weeks.

Automatic IETF Minutes