CELLAR

Summary

The CELLAR Working Group met to discuss progress on the Matroska specification (draft-ietf-cellar-matroska), the FFV1 v4 specification (draft-ietf-cellar-ffv1), reported errata for EBML (RFC 8794), guidance for Designated Experts, and the potential impact of AI-generated spam on IANA registries.

Key Discussion Points

FFV1 v4 Status

Jérôme Martinez reported that he has pending Pull Requests (PRs) for the FFV1 v4 specification (draft-ietf-cellar-ffv1) that he wants merged before making a new release.
Jérôme Martinez plans to send these PRs to GitHub over the upcoming weekend for review. Once merged by Michael (co-editor), Jérôme plans to publish the new draft by mid-October.

Errata for EBML (RFC 8794)

The group reviewed two reported errata for RFC 8794 (EBML).
Steve Lhomme noted that one of the errata corrects a clear bug regarding the maxOccurs attribute for attachments, which was incorrectly restricted when it should allow more than one.
Spencer Dawkins proposed, and the group agreed, to ask the Area Director to approve these errata.

Matroska Specification (draft-ietf-cellar-matroska)

Spencer Dawkins and Steve Lhomme discussed outstanding PRs and the upcoming transitions from version -20 to version -21 of draft-ietf-cellar-matroska.

PR 1127 and PR 1128: These include changing "codex" to "codecs" (plural) and changing uppercase "MUST/MUST NOT" to lowercase in the IANA Considerations section to avoid normative language.
PR 1130 (T35 Header): Steve Lhomme explained a change proposed by an external developer regarding the T35 header. Because real-world specifications are complex and bytes can be spread out, the text was reworded to refer to "all the bytes in the header" rather than "the first bytes." The group agreed to merge this PR.
"Should" vs. "Recommended": Spencer Dawkins highlighted a change from "should" to "recommended" made for clarity to match surrounding text.
- Note: In the session chat, Robert Sparks shared a link to RFC 2119 Section 3 to clarify that "SHOULD" and "RECOMMENDED" are synonyms under BCP 14 rules. Dave Rice acknowledged this clarification.

Designated Expert Guidance for Codec Registries

Spencer Dawkins raised a question on whether the Matroska specification should provide explicit examples of what constitutes a "usable specification" for legacy codecs (e.g., whether an open-source implementation like Libavcodec is sufficient).
Jérôme Martinez noted that the primary role of the registry is to prevent conflict and avoid reassigning codec IDs, making the exact quality of the reference specification secondary.
Steve Lhomme and Robert Sparks advised against adding highly specific reference criteria to the document, noting it could lead to complications. The group agreed to leave the current expert guidance text in draft-ietf-cellar-matroska as-is.

AI and IANA Registry Spam

Robert Sparks initiated a broader discussion regarding ongoing IETF-wide conversations about the impact of AI on registries. Specifically, there is concern that automated AI tools could flood First-Come, First-Served (FCFS) or Expert Review registries with requests, exhausting the available code space.
Steve Lhomme noted that while the Matroska codec ID space is virtually infinite, EBML Element ID registries contain highly limited, small ID spaces where exhaustion/spam is a real risk.
The group agreed that this issue should not be addressed within the current working group documents, and CELLAR will wait for eventual system-wide guidance from the IETF/IESG on managing registry spam.

Decisions and Action Items

Decisions

RFC 8794 Errata: The working group agreed that the reported errata are correct and should be moved to "Approved."
PR 1130: The working group agreed to merge PR 1130 into draft-ietf-cellar-matroska.
Designated Expert Guidance: The group decided not to add extra criteria defining "usable specifications" to the Matroska draft.

Action Items

Spencer Dawkins: Contact the Area Director to request the approval of the EBML (RFC 8794) errata.
Steve Lhomme: Merge PR 1130 and submit draft-ietf-cellar-matroska-21 incorporating the discussed PRs (1127, 1128, and 1130).
Spencer Dawkins: Update the summary email of Matroska document changes to reflect version -21 and send it to the working group mailing list for final confirmation.
Jérôme Martinez: Submit pending FFV1 v4 PRs to GitHub for review and subsequent merging by mid-October.

Automatic IETF Minutes