Markdown Version | Session Recording
Session Date/Time: 05 Mar 2025 15:00
CBOR
Summary
This interim meeting focused on the status of several CBOR Working Group documents, particularly cbor-cddl (CDE), cbor-packed (PACK), and cbor-edn-literals (EDN). A significant portion of the discussion revolved around the stalled progress of these documents and strategies to move them forward, especially in preparation for the upcoming IETF 119 meeting in Bangkok. Decisions were made regarding the initiation of a Working Group Last Call for cbor-cddl and the need for further discussion and alternative proposals for cbor-packed and cbor-edn-literals. The chairs also outlined a plan for the Bangkok meeting agenda and confirmed the continuation of interim meetings.
Key Discussion Points
-
Document Status Overview:
- RFC 9741 (CBOR Controls): Noted as complete and published, increasing CDDL's power. Thanks extended for Carsten's quick responses to O-48s during the process.
- Stalled Documents:
cbor-edn-literals,cbor-packed, andcbor-cddlare currently stalled, with dependencies and differing views preventing progress.
-
cbor-cddl(CDE):- The document has undergone significant editorial tweaking since its last working document stage over a year ago. No technical changes have occurred, but issues like the Nan representation and profile terminology have been addressed.
- A summary of ADR requirements by Lawrence has been proposed for inclusion in an appendix.
- A pull request for an old motivation text for implementers is open, but likely superseded by new section 2 additions.
- Decision: The chairs decided to initiate a Working Group Last Call (WGLC) for
cbor-cddl. - Action Item: Carsten Bormann committed to having a Pull Request (PR) for Lawrence's text by the weekend and a -09 draft available by the Sunday before Bangkok.
-
cbor-packed(PACK):- A hackathon in Dublin demonstrated successful implementation of the current draft.
- A key point of contention is the tag allocation strategy for extension points. Concerns were raised about the large number of tags consumed by the current proposal (beyond the "30-year outlook" guideline), contrasting with the initial intent for CBOR tags.
- An alternative proposal was briefly presented earlier, focusing on minimizing tag consumption.
- Action Item: Carsten Bormann to provide a list of drafts that reference
cbor-packedto understand their intended usage. - Action Item: Ori Pomerantz volunteered to write up an alternative proposal using essentially a single tag for comparison with the current approach.
- It was suggested that hackathon activity for
cbor-packedcould involve feeding new parameters into implementations to assess behavior and costs of different tag allocation strategies.
-
cbor-edn-literals(EDN):- This document aims to provide an extension point for diagnostic notation, particularly for CoRAL, and includes an ABNF for EDN.
- The document is stuck on discussions about implementation preferences, specifically regarding a single-layer versus two-layer ABNF structure and escaping mechanisms for app-specific strings.
- Rohan's proposal suggests a single-layer ABNF for app-strings, with extensibility via concatenation, removing default string escaping. Joe's proposal would allow escapes.
- A core requirement for EDN (derived from JSON) is the ability to flatten any EDN file to printable ASCII plus newlines for easier processing by older or limited systems. The impact of different ABNF approaches on this requirement, especially for non-ASCII characters, was debated.
- The ease of defining extensions versus the burden on implementers for performance and compliance with ABNF best practices was a central theme. The current document's two-layer ABNF requires implementers to understand a conversion or write tools, whereas a single-layer ABNF would be more aligned with common IETF practices.
- Decision: The discussion on
cbor-edn-literalswill continue and be a major focus at the Bangkok meeting.
-
IETF 119 Bangkok Meeting Planning:
- The CBOR WG has a 60-minute slot on Monday, with no known conflicts.
- Agenda Allocation:
cbor-edn-literals: At least 30 minutes, likely more, due to unresolved differences.cbor-packed: At least 10 minutes.cbor-dns: Martin Thompson requested 5-10 minutes.cbor-cddl: Expected to be largely handled via mailing list WGLC, but will be mentioned.
- Action Item: For
cbor-edn-literals, Rohan and Carsten (and potentially Joe) are requested to prepare presentations outlining the properties, pros, and cons of their respective proposals by Tuesday/Wednesday of the following week (the week before Bangkok) to facilitate pre-meeting discussion on the mailing list.
-
Interim Meetings:
- Decision: The working group will continue to hold interim meetings as they have been.
- The first interim after IETF 119 Bangkok is tentatively scheduled for April 2nd, 2024.
Decisions and Action Items
- Decision: Start Working Group Last Call (WGLC) for
cbor-cddl.- Action Item (Carsten Bormann): Provide a Pull Request for Lawrence's ADR text and a -09 draft of
cbor-cddlby the Sunday before IETF 119 Bangkok.
- Action Item (Carsten Bormann): Provide a Pull Request for Lawrence's ADR text and a -09 draft of
- Action Item (Carsten Bormann): Compile and share a list of drafts that reference
cbor-packed. - Action Item (Ori Pomerantz): Prepare a write-up of an alternative
cbor-packedproposal utilizing a single tag for comparison. - Action Item (Rohan Mahy, Carsten Bormann, and Joe Hildebrand if participating): Prepare presentations (or slides outlining key properties/pros/cons) of the
cbor-edn-literalsABNF proposals by Tuesday/Wednesday of the week before IETF 119 Bangkok to enable early mailing list discussion. - Decision: Continue holding interim meetings.
- Decision: Tentatively schedule the first interim after IETF 119 Bangkok for April 2nd, 2024.
Next Steps
- Review and discuss the
cbor-cddlWGLC on the mailing list. - Continue technical discussions on
cbor-packedandcbor-edn-literalson the mailing list, incorporating the newly requested materials. - Prepare for detailed discussions on
cbor-edn-literalsandcbor-packedduring the IETF 119 meeting in Bangkok, with substantial time allocated toedn-literals. - Consider specific hackathon activities for
cbor-packedto test different tag allocation parameters.