Session Date/Time: 12 Jun 2024 14:00

CBOR

Summary

The CBOR working group meeting covered the status of several in-progress documents, including CBOR Time Tag, CBOR Update Grammar, EDN Literals, CBOR M-Control, CBOR Pack, CBOR Deterministic Encoding (CDE), and CBOR Diagnostic Notation (DCBOR). Key discussions focused on the need for an additional last call for CBOR Time Tag due to new critical/elective key definitions, the parsing approach for appstr in EDN Literals, and the appropriate track for DCBOR. Decisions were made to initiate a re-last call for CBOR Time Tag, move CBOR M-Control towards submission, and initiate a working group last call for CDE after minor editorial fixes. Discussion on EDN Literals issues will continue on the mailing list.

Key Discussion Points

CBOR Time Tag (OFA 48):
- The document, previously in RFC Editor queue, was returned to ISG processing.
- Rohan Mahi's review identified a bug (missing ABNF registry entry for a map key) and a need for critical/elective versions of the time-scale key (minus one).
- New numbers (-13 and 13) were proposed for elective and critical time-scale keys.
- Francesca Baccaro (AD) confirmed that, despite previous approvals, this change warrants another working group last call to follow process, accepting a short delay.
- The question of whether the original -1 key should be marked "Legacy" was discussed, with consensus that it must retain backwards compatibility due to existing (though unknown) implementations.
CBOR Update Grammar:
- Completed IETF last call; awaiting AD review and ISG balloting. The AD confirmed comments were addressed and the document is ready to progress.
- A participant suggested that, like CBOR itself, CDDL updates could eventually be rolled into a single replacement RFC for implementer convenience, but recognized the current staged update approach.
EDN Literals:
- Completed IETF last call with seven open GitHub issues.
- A participant suggested defining EDN in its own RFC rather than as "extended diagnostic notation" to improve clarity for implementers, noting it has expanded beyond its initial scope. The chairs acknowledged the expansion but noted the difficulty of a title change after last call.
- GitHub Issue: Superfluous ABNF production: Christian Amsüss identified dead code in the ABNF (S production never reached). The AD indicated this might be acceptable if confirmed by implementations.
- GitHub Issue: Encoding indicators for long numbers: Discussion on adding specific encoding indicators for tag 2 (bigint) and tag 3 (bignegint). It was proposed to capture this as a W-defined extension point for future work. The document should clarify that long integers map to tags 2 and 3.
- GitHub Issue: appstr parsing: Discussion on the two-layer ABNF parsing approach for appstr (application-specific strings). A participant argued for a single, integrated ABNF layer for better clarity and ease of extension. Other participants and authors preferred the two-layer approach due to perceived implementation simplicity and reduced risk of breaking parsing for unknown extensions. This point requires further discussion on the mailing list.
- GitHub Issue: String concatenation: A proposal to reconsider/remove string concatenation (.. operator) to free up syntax for future features (e.g., +). This is a significant change late in the process, but could open up new possibilities.
- GitHub Issue: Guidance for tags 888/999: Not a technical change, but guidance for implementers on how to process these tags.
- Remaining Issues: Purely editorial (Genart review) or general guidance (security considerations), requiring some text additions.
CBOR M-Control:
- Received two observations post-IETF last call: clarification on .join definition (PR #5 merged) and strictness of base64 operators (PR #6 merged).
- -05 version submitted. The chairs anticipate these are minor enough not to require another working group last call.
CBOR Pack:
- Completed working group adoption call but is in "post-work last call exploration" due to a desire for more implementation feedback.
- An action item was made for authors to update additional resources/repository with links to existing implementations.
- CBOR Pack by Reference: This draft provides additional table setup mechanisms, extending CBOR Pack. The authors discussed whether it should be merged into the core CBOR Pack document. The sense was that reference mechanisms are core, but table setup mechanisms are often application-specific, suggesting they might remain separate. Implementation feedback on this draft is also desired.
CBOR Deterministic Encoding (CDE):
- Few open issues, mostly editorial tweaks.
- A quick poll of those present indicated 3 ready, 1 not ready, and 5 no opinion for a working group last call, suggesting sufficient readiness.
- A participant requested moving historical context to an appendix or adding a clear section for normative requirements.
- Discussion on the implicit nature of map sort order; it needs to be made more explicit in the document.
CBOR Diagnostic Notation (DCBOR):
- Discussion about whether to pursue this as a working group item or on the individual track (e.g., ISE stream).
- A chair expressed concern that WG adoption might lead to an over-cautious document diluted by caveats, whereas an independent track could better serve its specific use case (a CBOR application profile for integer/float issues).
- A participant requested that any objections or concerns be raised on the mailing list for open discussion and documentation.
CBOR Modules: Stable, but the treatment of sockets still needs prototyping and integration.
Numbers in CBOR Drafts: This informative draft is being revised; working group adoption was suggested to indicate WG interest, but no immediate action.
Formative Documents: DNS CBOR, Yang Standin (addressing textual info in Yang CBOR), and Yang Metadata (expanding Yang CBOR coverage) are in early stages, with Yang CBOR work moved to the CBOR WG.
Other Documents: Notable Tags, Deterministic Encoding Explainer, CDDL CSV, and three future CDDL work drafts are in various stages, with some potentially going to the ISE track.
Mimi Content Format: A recent I-D on instant messaging content using CBOR was highlighted, encouraging WG members to review and comment on the Mimi mailing list.

Decisions and Action Items

CBOR Time Tag: The working group will conduct another working group last call for the document to address the new critical/elective key definitions. This re-last call has already been requested by the AD.
CBOR M-Control: The document, with recent PRs merged, is deemed ready for submission unless objections are raised by the end of the week.
CBOR Pack: Authors are requested to update the additional resources/repository with links to existing implementations to gather more feedback.
CBOR Deterministic Encoding (CDE): A working group last call will be initiated. Editorial fixes (e.g., making map sort order explicit, restructuring history) are to be incorporated before or during this process.
CBOR Diagnostic Notation (DCBOR): A chair will initiate a mailing list thread to explicitly solicit objections and concerns regarding the document's progression on an independent stream versus working group adoption.
EDN Literals:
- Discussion on the appstr ABNF parsing approach will continue on the mailing list.
- Input is requested on all open GitHub issues, including the potential removal or change of string concatenation, and addressing editorial/security review comments.
Mimi Content Format: Working group participants are encouraged to review the Mimi content format document and provide comments on the Mimi mailing list.

Next Steps

Monitor the CBOR Time Tag re-last call.
Await AD action on CBOR Update Grammar.
Provide input on the EDN Literals GitHub issues, especially regarding appstr parsing and string concatenation.
Authors of CBOR Pack and Pack by Reference should seek more implementation feedback.
Prepare CDE for its working group last call, incorporating editorial suggestions.
Engage in the mailing list discussion regarding the appropriate track for DCBOR.
Review and comment on the Mimi content format I-D.