Minutes IETF102: lwig
Light-Weight Implementation Guidance
||Minutes IETF102: lwig
LWIG WG Meeting
IETF 102 - Montreal
Fairmont The Queen Elizabeth
Date: Friday, July 20, 2018, 11:50 - 13:20
Chairs: Zhen Cao, Mohit Sethi
AD: Suresh Krishnan
Presentation materials: https://datatracker.ietf.org/meeting/102/session/lwig
Meetecho for remote participants: http://www.meetecho.com/ietf102/lwig/
Etherpad for notes:
1. Administrative and Agenda Bashing (Chairs, 10 min)
Note Well, Note Takers, Jabber Scribes, Agenda Bashing
Minute takers: Francesca
Jabber Scribe: Rahul
Mohit announces 2 RFC being published (8387, 8352)
Slight change to agenda,: 4 won't be presented
2. Rahul: Neighbor Management Policy for 6LoWPAN (10 min)
Rahul presenting the slides
Mohit (no hat): think this is a very useful doc. Please note this is an
informational. Rahul : section in the draft "if you want to do neighbor
mgt, these are the things you will need"
3. Carlos: TCP Usage Guidance in the Internet of Things (IoT) (10 min)
Presenting the status and updates to the draft
Rahul: I think the information on socket interface in OpenWSN in red is
incorrect; BSD-style socket interface is not possible
Carlos: Ok, we will check
Markku: Thank you for doing this, useful work. Needs to be more specific
regarding turning off Delay ACKs (global, per interface, ...). Another
example is ECN, the current description of the RTO behavior is not correct.
Will do a full review
Carlos: Thanks. About the delay technology, we already have some text
Markku: still need to be more specific
Mohit (no hat): Could you post this draft ?
Zhen: could be good to have the discussion in lwig and tcp wg
Carlos: I think we had the discussion and we would have the last call in
both working groups
Rahul: Is any of these stacks already supporting ECN?
Carlos: No, not yet.
4. Carsten: Virtual reassembly buffers in 6LoWPAN (10 min)
5. Olaf: CoAP Implementation Guidance (10 min)
Olaf (remote) presenting the update to the draft
Carsten: as a co-author question for the wg, do people perceive a rush to
get this published? In CoRE we had a doc like this who waited 5 years to
get published. Same here or not? Matthias: also as co-author, no reviews
coming in. before wglc I feel like reviews don't happen Mohit (no hat):
good idea but need an editor to make sure that someone brings in all the
comments Carsten: maybe Matthias is right
Zhen: lack of review is ?? . (missed it)
Matthias: plugtest: we got something from F-interop. some issues that makes
sense to document, others maybe not
Hannes: chairs maybe should organize interop on features including CoAP
over TCP, which could be fed into this doc. Comments we see is
clarifications people want
Mohit: we can talk to CoRE chair and organise something
Carsten: we should talk to F-Interop
Hannes: I think F-Interop have a different goal, come up with an automated
test.we don't need that in the first place Mohit: some interop is better
than not any. Carsten: need to do more of those. Nice thing of interop is
people are sitting together and talk to each other. Maybe organizing
something in Prague time frame would be good Mohit: sounds good
6. Carsten/Ari: Terminology for Constrained-Node Networks (10 min)
Carsten presenting the slides
Hannes: (J-group classes slides) (...) Virtualization imo it provides
another layer of protection. enhance security capabilities.
Carsten: would like to get feedback from the wg
Hannes: are you planning to talk about requirements (devices)? won't need
to be super accurate, but give a rough idea
Carsten: good idea start collecting expected burdens for classes, that
would fit very well with this document
7. Daniel: Minimal ESP (10 min)
Daniel presenting the slides
How many people have read this doc? ~10
in favor of adopting: some hum
against: no hum
Rene: did you fix the algorithm option?
Daniel: no, we described how to select an algorithm
8. Rene: Alternative Curve Representations (10 min)
Rene presenting the slides
Carsten: question to your last question: yes. It is really hard to do
crypto on small devices.
Sean Turner: section 5 is the most interesting section. CURDLE is looking
at curves and deprecating and having new things. Think about that. Second
point: this is mostly about the identifiers, the chip already has the code.
Rene: as an implementer you could take a curve and do a very simple
Sean: trying to understand the driving motivation for this
Rene: if you have a small device and you implement ECDSA and Ed-25519 (...)
in 6lo they have may options to do things and not all devices speak all
Carsten: real example one P-256 and Ed25519, you have to have both
Anybody objecting to this becoming a wg doc?
Daniel: as co-chair of CURDLE, I am fine if you ask for review in Curdle,
you would do that very fast, since the wg is not very active
Rene: what is the output of Curdle?
Daniel: deprecating and providing points for new alg. But this would be
just for the review.
Hum for adoption: hum for adoption, no hum for no adoption
9. Francesca: Comparison of CoAP Security Protocols (10 min)