Markdown Version | Session Recording

Session Date/Time: 30 Apr 2025 14:00

CBOR

Summary

The CBOR interim meeting addressed the status of three key documents: CBOR Diagnostic and Example (CDE), CBOR Packed (CBOR-PS), and CBOR Extended Diagnostic Notation (EDN Literal). For CDE, ongoing Working Group Last Call (WGLC) comments and open GitHub issues need resolution. Discussion around CBOR Packed focused on setting "tunable" efficiency parameters (A, B, C) and a new alternative proposal was deferred. For EDN Literal, the discussion centered on the appropriate layer for expressing advanced functionalities like hashes and encryption, with a new Pull Request for review.

Key Discussion Points

• CBOR Diagnostic and Example (CDE)
  • No significant progress was reported as the working group is awaiting comments from the Working Group Last Call.
  • Several open issues on GitHub were noted, which need to be addressed and discussed on the mailing list. It was clarified that there needs to be active discussion on these issues, not just open tickets.
  • The group agreed on the need to "put a box around" the comment period and resolve the outstanding issues.
• CBOR Packed (CBOR-PS)
  • Karsten presented on Pull Request #24, which proposes a "radical solution" for reducing resource consumption in CBOR Packed, while retaining three specific efficiency improvements as tunable parameters (A, B, C).
  • A discussion arose regarding the rationale for these tunable parameters (A, B, C) and whether a "truly radical" reduction (A=B=C=0) would meet the requirements. There was a call for more quantitative or requirements-based justification rather than subjective terms.
  • Participants were encouraged to run experiments with their specific applications and contribute their views on reasonable values for A, B, and C to the mailing list.
  • Vadim presented an alternative proposal for CBOR compression, which uses a different approach involving internal escaping and potentially reserved simple values, aiming for high compression effectiveness similar to DNS name compression.
  • Karsten expressed concern that Vadim's proposal seemed like a "whole new work item" requiring a complete Internet Draft, as it involves concepts that could alter CBOR invariants (e.g., reinterpreting tag numbers) and require a fundamentally different decoder.
  • The group briefly touched upon the definition of "compression" in the context of CBOR Packed, clarifying it's about exploiting redundancy and enabling shared representations without necessarily requiring full decompression before operation.
• CBOR Extended Diagnostic Notation (EDN Literal)
  • Discussion initiated on the appropriate layer for expressing functionalities like cryptographic hashes, signatures, or AAD encryption in specifications.
  • Lawrence suggested that such semantics are better handled by CDDL (CBOR Data Definition Language) through new controls or by pre-processors using existing EDN features like comments and ..., rather than adding new application extensions directly to EDN. He emphasized maintaining a simple syntax layer for EDN, similar to JSON's success.
  • Karsten acknowledged the value of CDDL for defining semantics but also highlighted the utility of simple EDN app extensions for writing concise examples in specifications. He noted the potential for future symmetry between CDDL operators and EDN mechanisms.
  • A new Pull Request 77 for EDN Literal was mentioned, and participants were asked to review and comment.

Decisions and Action Items

• CDE: The working group will continue discussion on the mailing list to resolve open GitHub issues. The next meeting will explicitly go through these issues.
• CBOR Packed:
  • Participants are requested to provide input on desirable values for the tunable parameters A, B, and C (from Pull Request #24) on the mailing list, ideally backed by application-specific experiments or rationales.
  • Vadim's alternative compression proposal is deferred ("parked") until the working group resolves the current set of documents (CDE, CBOR Packed, and EDN Literal).
• EDN Literal: Discussion on the extensibility model (app extensions vs. CDDL controls) will continue on the mailing list. All participants are encouraged to review and comment on Pull Request 77.

Next Steps

• Mailing List Activity: Active discussion is expected on the mailing list for:
  • Resolving open issues for CDE.
  • Providing input and rationale for A, B, C values in CBOR Packed.
  • Continuing the discussion on the EDN Literal extensibility model and reviewing Pull Request 77.
• Next Meeting: The next interim meeting will specifically address the open CDE issues and continue with the EDN Literal discussion.