[Cbor] Minutes from 4th webex call

[Francesca Palombini <[e-mail]>]

Hi all,

Here are the notes from today’s interim meeting, also uploaded on the datatracker: https://datatracker.ietf.org/group/cbor/meetings/
Participants, as always, please let me know if anything was not captured correctly.

Please note that the next call (in two weeks) is cancelled, and we will meet directly in Bangkok.

Thanks,
Francesca

--

CBOR WG Conference Call
Wednesday, Oct 17 2018, 15:00 - 16:00 UTC
Chairs: Barry Leiba, Francesca Palombini

Minute takers: Francesca

Participants:
* Barry
* Carsten Bormann
* Michael Richardson
* Paul Hoffman
* Francesca
* Laurence Lundblade

Hackathon 103:
    Barry: are we going to have an interop?
    Carsten: if we have issues where an interop would help
    Barry: Not necessarily, but rather for visibility
    AP Barry: presence at the Hackathon, at the wishi table, come and join us

CDDL status:
    Carsten: IANA questions
                                * CDDL control operators are new registry
                                * should the entries be alphebatized
                Barry: why not, but it does not really matter
                Carsten: if it makes more logic sense, having them in another order
                2 reviews to fix before IETF103
                Waiting for writeup from Alexey
                AP Barry: ping Alexey
                Telechat in december hopefully
                For Bangkok we can summarize the changes from reviews

CBOR Bis
                Carsten: any substential feedback from Jeffrey we need to discuss in Bangkok
                Paul: one major issue is key equivalence, a number of minor issues. Pull requests waiting to be fixed to be mergeable.
                Carsten: first commit on the issue from Laurence 6 days ago. Most of the work is done but we have to agree that we agree.

                map key equivalence commit
                Carsten going through https://github.com/cbor-wg/CBORbis/commit/d2ba777d54faa649c8e6b30c09e19ac898426be8
                                - generic definition
                                - preferred encoding for floating point values (all the time not just for when they are keys)
                                - data model use in equivalence of keys section
                                - new text about preferred serialization: started from Laurence text but editorially different
                Michael: doesn't like this non-binary interoperability.
                Laurence: looking for clear rule that if you follow you always are going to be interoperable
                Carsten: maybe can be rephrased in "some protocol create the opportunity to have that"
                Michael: this is not under implementation section. "protocol designers may want to use constrained decoders, they need to call that out, whenever they make use of shortcut that does not make use of preferred serialization"
                Laurence: most interested in indefinite length.
                Carsten: make the role of the protocol more clear.
                Laurence, Michael: yes
                Michael: if we don't, people will make bad assumptions
                Carsten: Ok, thanks

                Laurence: PR for dates and for shortest floating point
                Carsten: floating point is the commit above, the other still on the todo list
                Laurence: reference IEEE754
                Carsten: yes we should

                Laurence: precision for floating point
                Carsten points to l. 1270-1278 in commit above
                Carsten: I think putting information in here would be wrong. Important part is that if you choose an encoding that loses precision, that should be something that is explicit, desired buy the application, but the other direction is not very useful
                Laurence: how do you decide if something can make been shorter or not?
                Carsten: line 1271
                Laurence: we should interop test
                Carsten: Hackathon 103

                Michael: suggestion: we should have some text about "cbor based protocol is supposed/expected to use all preferred serialization"
                Carsten: yes we can add that

                Carsten: next steps: wait on Jeffrey's feedback before including his requests. Try to get Laurence's in before drafts cutoff.

                AP chairs: Agenda for Bangkok meeting to be requested in the mailing list

                Array tag: Carsten to check that the draft was updated and then adopt

                Next conference call is cancelled and we meet in Bangkok