interim-2021-cbor-03
2021-02-10 15:00 - 16:00 UTC
Chairs: Francesca Palombini, Christian Amsüss
Recordings: https://youtu.be/Jn9UHzzOk1A
Attendees:
Francesca doing introuctions; no changes to the agenda.
FP: Two weeks no IPR response from Sean; next steps?
CB: Last time contacted via EBay (but still alive ;-) ). His original parts are, in most parts, gone into the other document during split. Wouldn’t be entirely wrong to move him to contributors.
CA: Given web address is nonresponsive.
FP to BL: Typical procecedure?
BL: On IPR, move to contributors doesn’t help; a significant contributor must confirm. Personal knowledge of “no IPR involved” would be OK, otherwise trouble in AUTH48. Then the AD can approve for him as last resort. For AUTH48, moving people that can’t be reached makes things easier. For Shepherd writeup a “we believe that BSP79 compliance is here but don’t have confirmation” is OK (→CA merge to text).
FP: Wait one more week for responses on other channels, then move to contribs and finish shepherd review.
CB: Example from 2017 hackathon (W3C WoT) now included (just using prefix mechanism; partially hand-crafted). No plan for circumfix expansion yet. Also: template draft for LPWAN. Best possible outcome would be an understanding that this is an optional ingredient. The example shows why the circumfix proposal would be useful.
Numbers allocated, stuck to 1/4 space for suffixes.
CA: Is dictionary content to be agreed on verbatim? (Could help with templating and circumfix expansion).
CB: Currently mechanical construction, should stick to that. Other ways of building them may be more dynamic.
CB: Roadmap: Implementation in ~10 days, play with it.
FP: one more meeting before IETF110
MR: Tagging for on-disk storage [showing slides; notes don’t cover what it says there]
Prefix tag to original item, or CBOR sequence?
Latter, because it’s easy to parse and have the rest free of tags.
CA: 55799 says “what’s inside is tagged” (that’s OK) but also “this is a CBOR file” and that’s inaccurate with the stream thing
MR: Adopt and then change it.
CB: 55799 has arithmetic in it that makes it non-Unicode in any encoding, just make sure the new tag has the same property.
CB: For people doing a CBOR file (and not seq), would still like convention spelled out for how to do this there. Still go for 32bit tag, but just needs wording about the encoding. But have both, and keep recommendation for sequence.
FP: Question on scope for WG was asked; own opinion yes, would like to hear more people’s statements both ways here and then confirm on ML.
(+1 from PY; “where else?”)
FP: Mark as BCP?
MR: We can allocate (2+ tag) with BCP, right?
CB: It’s first-come-first-served.
MR: So come up with number, register pointing to I-D, then go ahead. But nothing standards-level.
FP: Adoption, then details.
MR: [see slides again]
Original 261 definition link: https://github.com/toravir/CBOR-Tag-Specs/blob/master/networkPrefix.md
Concerns about forcing trailing zeros, and byte alignment.
FP: CB as expert, opinion on early alloc?
CB: It’s Specification Required only.
MR: Only reason against notable-tags would be if early allocation wouldn’t be an option. So whatever is fastest.
CB: Own document seems to be least complex.
FP: Yes, keep it as is.
MR: Might be timely for 6MAN who have CBOR encoded RA option (by Ole; but not adopted yet: draft-troan-6man-universal-ra-option-04 ). This may be interesting to have there. Encoding the V6 prefix in short may help.
CB: Still makes sense to have the tag defined even if it’s not sent on the wire.
FP: Looks like in scope; voices against adoption? [silence] Adoption call on ML.
CB: If adopted, consider what we do with OID to make it commute with array.
MR: It’s unambiguous but needs some peeking.
FP: Adoption calls will come, please reply to thread even if you voiced in the call!
→CA: Send adoption calls.
CB: W3C DID spec has some CBOR content; not sure what yet. May be worth looking at.
→ CA look through it.