Session Date/Time: 25 Jan 2022 20:00

CELLAR

Summary

The CELLAR WG meeting discussed the status of its drafts, focusing on EBML, Matroska, and FLAC. Key topics included plans for an EBML errata update and its potential advancement to Internet Standard, the progress on the Matroska specification including IANA registry generation, and the structure of the FLAC document with proposed appendices for implementation guidance. FFV1 had no new updates.

Key Discussion Points

Attendees: Michael, Jerome, Martin, Steve (Chair). Christian joined as a lurker.
Previous Minutes: Minutes from the previous meeting were reviewed and accepted without objection.
Note Well and Code of Conduct: The IETF Note Well and Code of Conduct (BCP 25 and 54) were reviewed.
Draft Status Overview:
- FFV1 was reposted a few days after the last meeting.
- The group acknowledged the need for continued progress on Matroska and other drafts.
EBML Update and Advancement:
- Errata: Discussion centered on updating the EBML specification to include errata discovered during Matroska development and reported by implementers since its RFC publication in 2020.
- Dependency: It was reiterated that work on the EBML update, particularly any errata, would not proceed until the Matroska specification is finalized.
- Internet Standard Advancement: The possibility of advancing EBML from Proposed Standard to Internet Standard was discussed. Criteria for advancement include being a Proposed Standard for two years and having at least two interoperable implementations.
- Definition of Interoperable Implementations: Consensus was that for EBML, interoperable implementations would specifically mean two implementations of Matroska (which uses EBML), not just two disparate uses of EBML. This reinforces the dependency on Matroska's finalization.
- Errata vs. Advancement: The group considered whether discovering further errata after advancement would be an issue. It was agreed to defer the decision on requesting Internet Standard status until after any errata are incorporated into a revised Proposed Standard.
- Feature Support: The group discussed how to handle features not fully supported across all existing EBML/Matroska implementations (e.g., "unknown size"). The document might note caveats for such features, or if a feature is widely unused, it could be removed. Given Matroska's age and existing implementations (especially in hardware), a subset of features being widely used might be noted as the "fully interoperable" core.
Matroska Status:
- Recent Merges: Jerome reported merging several changes, including a rewrite of time definitions (clarifying different time bases) and the removal of sample-precise timestamps due to backward compatibility issues.
- ChapterTimeEnd: The mandatory status of ChapterTimeEnd was clarified; it is now mandatory only for the lowest level of nested chapters.
- IANA Registry Generation: A tool has been developed to generate the IANA registry from the XML source, similar to the EBML one. It includes actual values but does not currently fill gaps between IDs (marking them as "unassigned"). This tool's output was reviewed.
- Next Steps for Matroska: Jerome identified 9 open pull requests (PRs) linked to remaining issues for Matroska. He indicated the document would be ready for wider review once these PRs are merged. Key PRs needing merger include:
  - PR 521 (Matroska timestamps/time bases).
  - PR 115 (fonts).
  - The IANA generation PR (578) could be merged immediately.
FLAC Status:
- Appendices: Martin discussed two new appendices for the FLAC document (PR 120): "Numerical Considerations" and "Examples".
- Appropriateness: There was a discussion about the appropriateness of such content in an IETF standard. The group generally agreed that implementation advice and worked examples, especially in non-normative appendices, are valuable for implementers and can help clarify the specification.
- Numerical Considerations: This appendix provides guidance on data types for arithmetic operations (e.g., using 64-bit signed data types for safety) and considerations for optimizing performance (e.g., for DSPs or portable devices).
- Examples: These provide step-by-step decoding examples with hexadecimal interpretations, helping implementers verify their code. It was stressed that these examples must be robust and explicitly marked as informational, as any error could mislead implementers. Martin noted that preparing these examples helped identify areas needing more clarification in the main specification.
- Terminology: Jerome suggested that some terminology introduced in the "Numerical Considerations" appendix might be useful to define earlier in the document (e.g., in an introduction).
FFV1 Status: No new technical updates or issues were discussed.

Decisions and Action Items

Decision: Matroska IANA registry generation PR (578) to be merged.
Decision: Defer the decision on advancing EBML to Internet Standard until Matroska is finalized and EBML errata are incorporated.
Action Item: Jerome to merge Matroska PRs 521 (timestamps) and 115 (fonts), and PR 578 (IANA generation) soon.
Action Item: Martin to read the FLAC "Numerical Considerations" appendix again and suggest any terms that should be moved to or defined in the document's introduction.
Action Item: Martin (ktmf01) to merge FLAC PRs and post a new version of the FLAC document once comfortable, especially after internal review of the examples' accuracy.
Action Item: Steve granted Martin (ktmf01) write access to the repository to merge FLAC PRs.
Action Item: Martin to review older FLAC issues and close any that are obsolete or no longer relevant.

Next Steps

Jerome will continue to merge Matroska PRs, aiming for a new document version after the key PRs (521, 115, 578) are integrated.
Martin will finalize and post a new version of the FLAC document.
The group will continue to monitor the status of EBML, with an update pending Matroska's completion.
The next meeting is scheduled for the 22nd.