Markdown Version | Session Recording

Session Date/Time: 14 Jun 2023 14:00

CBOR

Summary

This interim meeting focused on planning for IETF 117 and reviewing the status of several key CBOR-related drafts. Significant progress was reported on the Time Tag document and the CBOR Packed specification. The working group formally adopted the CDDL 2.0 roadmap documents (Grammar Update, Control Operators, and Modules) and the EDN Literals draft. A proposal for splitting the deterministic CBOR discussion into two documents was presented and received positive initial feedback. A substantial part of the meeting was dedicated to clarifying issues around IANA tag registration, particularly concerning the Gordian Envelope draft, with a sense that while the core deterministic CBOR (DC-BOR) work would proceed, the broader Gordian envelope itself might not become a CBOR WG item at this time, pending further community interest and technical clarity.

Key Discussion Points

• Planning for IETF 117: The chairs outlined a planning outlook, noting the upcoming internet draft deadline and the IETF week. The goal is to review the status of various documents and plan activities for IETF 117.
• Time Tag Document: This document, defining CBOR structures for timestamps and durations, has been stable and received tag allocations. It is awaiting the completion of the SEDATE RFC last call (scheduled to close the next day) before being submitted to the ISG for publication. No significant impacts from SEDATE are anticipated.
• CBOR Packed: This document passed working group last call last year. It defines tags for referencing dictionary information. A primary outstanding item is gaining more implementation and user experience with the table setup, specifically concerning tag 113. The suggestion was made to potentially merge the default shared and argument table setups. An interim meeting in the coming weeks was suggested to discuss this further before IETF 117.
• CDDL 2.0 Documents (Grammar Update, Control Operators, Modules):
  • The call for adoption for the CDDL 2.0 Grammar Update, Control Operators, and Modules documents officially closed successfully.
  • The Grammar Update (a fix document for RFC 8610) and Control Operators (exercising RFC 8610 extension points) are expected to proceed quickly to working group last call.
  • The Modules document, while implemented, needs more feedback and independent implementations. Discussion is expected to continue beyond IETF 117.
• Diagnostic Notation (EDN Literals): This document has been adopted. A need was identified to potentially define ABNF for the CBOR Diagnostic Notation (EDN). While RFC 7049 stated EDN was not meant to be parsed, the decade of coexistence with actual CBOR encoding suggests less risk of confusion now. A -01 draft including ABNF could be proposed for discussion at IETF 117.
• Deterministic CBOR and Numeric Reduction (DC-BOR):
  • A proposal was made to split this into two documents:
    1. An informational document on standard deterministic encoding, explaining choices and profiles allowed by RFC 8949 Section 4.2.
    2. A document defining the DC-BOR numeric reduction, building on the standard deterministic encoding.
  • This approach received positive feedback. Christopher Allen offered to contribute implementation experience and test cases.
  • The objective is to have initial "-00" drafts ready for discussion at IETF 117.
  • The question of CDDL support for numeric reduction was raised. It was suggested this might best be handled in a separate document due to the complexity of representing models that mix components with and without numeric reduction.
• Other Active Work:
  • CBOR Draft Numbers: An informational document on best practices for allocating code points during development. Its effectiveness needs to be assessed.
  • DNS CBOR: Identified as an important application, also serving as a customer for CBOR Packed.
  • CDDL CSV: Work on this is deemed useful and ongoing.
  • Corpus of CDDL Models: A project to collect and package useful CDDL models from RFCs for easier referencing.
• IANA Tag Registration Process & Gordian Envelope:
  • Christopher Allen highlighted difficulties encountered when registering tags for the Gordian Envelope draft, specifically for the "Specification Required" range (24-32767), where IANA rejected requests despite the specification's guidance (RFC 8949) not requiring an RFC-track document. This process was experienced as a "pain."
  • Karsten outlined the role of Designated Experts (currently Karsten and Christian) in reviewing "Specification Required" requests, referencing RFC 8126 (Section 4.6). He explained that experts apply "frugality" due to the long-term nature of tag allocations and that specifications are thoroughly reviewed, which takes time. Confusion arose because the Gordian draft was presented as both complete and potentially input to an IETF process.
  • Specific tags in draft-mcnally-envelope-02 were discussed:
    • Tags like 204 (digest, e.g., SHA-256), 205 (encrypted, e.g., Chacha20/Poly1305), and 206 (compressed, e.g., Deflate) embody specific algorithms. Karsten noted these raise concerns about algorithm agility and are more suitable for dedicated registries (RFC 1954 for hashes, RFC 1952/1953 for encryption) or "3-byte" first-come, first-served tags. A lack of a CBOR-specific compression registry was noted.
    • Tag 202 ("known value") was unclear and identified as a "big hole" in the specification.
  • Christopher clarified that Gordian tags were designed for agility with default algorithms specified, but this might not have been clearly expressed. He reiterated the desire for the CBOR WG to take up Gordian envelope work, but acknowledged the lack of broad community interest.
  • The chairs indicated they would assess working group interest for Gordian as a work item. Karsten expressed personal interest in graph structures in constrained applications.
  • Given the complexity and time constraints, it was suggested that further discussions on Gordian, especially regarding tag allocation issues, should continue offline between the Designated Experts and the authors. Christopher suggested organizing side meetings at IETF 117.

Decisions and Action Items

• WG Adoption: The working group officially adopted draft-ietf-cbor-cddl-grammar-update, draft-ietf-cbor-cddl-control-operators, draft-ietf-cbor-cddl-modules, and draft-ietf-cbor-edn-literals.
  • Action Item: Chairs to submit these adopted documents as IETF drafts.
• Deterministic CBOR (DC-BOR):
  • Decision: Proceed with splitting the deterministic CBOR discussion into two new documents: one informational on standard deterministic encoding and one on DC-BOR numeric reduction.
  • Action Item: Karsten and Christopher Allen to collaborate on creating initial "-00" drafts for these two documents.
• EDN Literals (ABNF for Diagnostic Notation):
  • Decision: The working group is open to considering defining an ABNF for the CBOR diagnostic notation.
  • Action Item: An initial "-01" draft of draft-ietf-cbor-edn-literals including the ABNF will be prepared for discussion.
• Gordian Envelope Tag Registration:
  • Decision: Further discussion on the specific Gordian tags and the IANA registration process issues will continue offline between the Designated Experts and the Gordian authors.
  • Action Item: Christopher Allen to review RFC 8126 Section 4.6 for clarification on "Specification Required" expert review criteria.

Next Steps

• Time Tag Document: Await completion of SEDATE last call, then ISG submission.
• CBOR Packed: Consider scheduling an interim meeting (e.g., in two weeks) to specifically address implementation feedback on table setup (tag 113) and discuss its path forward for IETF 117.
• CDDL 2.0 Modules: Continue to gather feedback and seek multiple independent implementations; further discussion expected beyond IETF 117.
• DC-BOR Documents: Aim to have "-00" drafts ready for discussion at IETF 117.
• Gordian Envelope: Explore organizing side meetings at IETF 117 for further discussion, particularly if broader working group interest emerges. The Designated Experts will continue offline discussions with the authors regarding tag allocations.
• IETF 117 Agenda: Chairs will finalize the preliminary agenda for the CBOR WG session at IETF 117, allocating time for the discussed topics (e.g., 15 minutes for Gordian if pursued).