Session Date/Time: 10 May 2023 14:00

CORE

Summary

The CORE Working Group held an interim meeting primarily to discuss recent changes and progress on draft-ietf-core-href. Karsten Bormann presented the proposed changes for the upcoming -13 version of the draft, focusing on refinements to CRI semantics, the introduction of "unprocessable CRIs," and constraints on indefinite length encodings in extensions. The discussion also touched upon preparations for a Working Group Last Call and a follow-up on IANA related requests from other organizations.

Key Discussion Points

draft-ietf-core-href-13 Changes Overview: Karsten Bormann presented a diff of changes intended for the -13 submission.
CRI Semantics (Section 5.1):
- Discussion on the representation of an "empty path" in absolute CRIs. To align with CI reference semantics, it was proposed and generally accepted to require an empty array for an empty path, consuming one byte in CBOR, rather than implicitly allowing a "null" concept for an empty path.
- The explicit ruling out of an entirely empty CRI reference was discussed, as it does not appear particularly useful for interchange.
- It was clarified that "null" is not explicitly forbidden for paths but rather must be represented as an empty array.
CRI Validity and Unprocessable CRIs (PR 62 / Section 6):
- The concept of an "unprocessable CRI" was introduced. This refers to a CRI that might be syntactically well-formed but cannot be processed by a specific implementation due to, for example, unsupported extensions or features.
- It was emphasized that the duty of checking validity lies with recipients that rely on that validity. Intermediate nodes that do not process the detailed information might still be able to handle the CRI (e.g., for reference resolution) even if they don't understand specific features (like percent-encoded text).
- Producers bear the responsibility to create interoperable CRIs.
- Unprocessable CRIs were compared to "Not-a-Number" in floating-point representations, with the nuance that CRIs with identical serializations should be treated as identical. Implementations should handle unprocessable CRIs robustly without necessarily aborting processing.
Indefinite Length Items in Extensions:
- A strong suggestion was made to mandate that CRI extensions must not use indefinite length CBOR encodings. This simplifies processing and skipping of unknown extensions.
- A sense of those present indicated support for this constraint, with Christian confirming agreement in the chat.
- The broader question of prohibiting indefinite length encoding for CRIs in general (not just extensions) was briefly discussed, acknowledging that if a CRI is embedded within a larger CBOR structure that uses indefinite length for other reasons, it might be challenging to impose this constraint solely on the CRI part. The focus remains on CRIs when they are handled "on their own."
Normalization of Dots in Hostnames:
- It was clarified that dots (.) in hostname elements are not reserved and that normalization processes will convert any percent-encoded dots (%2E) back to literal dots.
Contributors: Christian Amsüss was added to the list of contributors. The chairs will follow up with Klaus Hartung regarding his inclusion.
IANA Discussion Follow-up: A request was made to follow up on previous IANA registry proposals from OneM2M and Global Platform. The chairs will work to identify appropriate contact persons from these organizations to present their requests for further discussion, potentially in an informal meeting or future interim.

Decisions and Action Items

Decision: For empty paths in absolute CRIs, an empty CBOR array ([]) MUST be used for consistency with CI reference semantics. (00:10:00)
Decision: An entirely empty CRI reference is explicitly ruled out for interchange as it offers no useful semantics. (00:12:01)
Decision (Strong Consensus): CRI extensions MUST NOT use indefinite length CBOR encoding to simplify processing and extensibility. A general recommendation against indefinite length for CRIs when handled independently was also noted. (00:20:02, 00:26:01)
Decision: Dots (.) in CRI hostname elements are not reserved; normalization processes will unescape percent-encoded dots. (00:22:00)
Action Item: Karsten Bormann to finalize draft-ietf-core-href-13 by addressing remaining parts of Issue 44 (specifically, defining "minimal representation") and incorporating editorial fixes, with a goal to submit it this week. (00:24:12)
Action Item: Karsten Bormann and Chairs to prepare draft-ietf-core-href-13 for Working Group Last Call. (00:24:12)
Action Item: Christian Amsüss to review draft-ietf-core-href-13 during the Working Group Last Call. (00:28:02)
Action Item: Karsten Bormann to follow up with Klaus Hartung regarding his inclusion in the contributor list for the href draft. (00:30:02)
Action Item: Martin Thomson and Karsten Bormann (Chairs) to identify contact persons from OneM2M and Global Platform regarding their IANA requests and schedule a discussion session (informal or interim) to clarify these requests. (00:32:00)

Next Steps

Submission of draft-ietf-core-href-13 and initiation of Working Group Last Call.
Scheduling a discussion with representatives from OneM2M and Global Platform regarding their IANA registry update requests.
Chairs will call for agenda items for the next interim meeting, potentially including progress on seed or other CORE WG drafts.