CBOR

Session: CBOR Working Group Interim Meeting Date: May 2026 (Interim-2026-cbor-09) Chairs: Barry Leiba Note Taker: Technical Writer (IETF)

Summary

The session focused on advancing draft-ietf-cbor-edn-literals toward Working Group Last Call (WGLC), discussing a potential renaming of the specification to address naming inconsistencies and collisions. The group also discussed the technical challenges of partial implementations and serialization constraints in draft-ietf-cbor-serialization, specifically regarding how constrained devices handle complex CBOR data models.

Key Discussion Points

draft-ietf-cbor-edn-literals Status and Renaming

WGLC Planning: Barry Leiba outlined the goal to move the document to WGLC on Monday for a 10-day period. This push is intended to determine if rough consensus exists or if the document needs significant re-evaluation.
Current Updates: Carsten Bormann reported that version -24 has no open issues. A new version (-25) is planned to fix editorial notes in the abstract and integrate PR 102 (shifting simple syntax range checks from semantic to syntactic).
Naming Discussion: Carsten Bormann presented a "cacophony of terms" regarding "Extended Diagnostic Notation (EDN)." Problems identified include:
- The term "Extended" is often redundant or vague in protocol names.
- The acronym "EDN" conflicts with the Clojure programming language’s "Extensible Data Notation."
- File extensions like .diag are not easily searchable.
Proposals for Renaming:
- Option 3: "CBOR Diagnostic Notation" (CDN).
- Option 4: "Concise Diagnostic Notation" (CDN).
Discussion on Naming: Laurence Lundblade expressed support for CDN, noting it aligns with CDDL. Mikolai Gütschow suggested the abbreviation should be unique (e.g., CEDN or CBORDN) to avoid confusion with Content Delivery Networks. Carsten Bormann expressed a preference for "Concise Diagnostic Notation" (Option 4).

Implementation Progress

Laurence Lundblade emphasized the need for "thorough" implementations rather than prototypes to validate the specification, particularly regarding backticks and complex features.
Carsten Bormann confirmed two existing implementations: one in Ruby (Bormann) and one in JavaScript/TypeScript (Joe Hildebrand). Вадим Гончаров noted that implementation is relatively trivial using standard lexer/parser tools (Flex/Bison).

Partial Implementations and Serialization Constraints

Presentation: EDN, Partial Implementations
Data Model vs. Serialization: Carsten Bormann argued for distinguishing between data model constraints (e.g., a device not supporting floating point or bignums) and serialization constraints (e.g., definite-length only).
The "Sweet Spot": Laurence Lundblade argued that the distinction is largely academic for most implementers. He suggested identifying a "sweet spot" subset of CBOR—similar to JSON’s scope (64-bit integers, strings, arrays, maps with simple keys, definite length)—that covers 80% of use cases without the cost of full CBOR decoding.
Complexity Concerns: Laurence Lundblade stated that full 8949 compliance is extremely costly for C-based constrained devices, requiring significant effort to handle complex map keys and indefinite lengths.
Interoperability: Barry Leiba noted the standard protocol paradigm: senders send what they need, but receivers must generally support the full specification to avoid errors. The group debated whether CBOR should formalize "profiles" of partial implementation, though Carsten Bormann cautioned against the complexity of formalizing such levels.

Decisions and Action Items

Renaming: Carsten Bormann will update draft-ietf-cbor-edn-literals to version -25, adopting the name "Concise Diagnostic Notation" (CDN). This naming change will be tested for consensus during the WGLC.
Editorial: Carsten Bormann to include boilerplate text for editorial conventions in version -25.

Next Steps

WGLC: Barry Leiba to initiate a 10-day Working Group Last Call for draft-ietf-cbor-edn-literals (to be renamed) starting Monday, pending the submission of the updated draft.
Mailing List: Discussion regarding partial implementations and the "sweet spot" subset for constrained devices will continue on the list to inform draft-ietf-cbor-serialization.

Automatic IETF Minutes