(I just realized that I forgot to CC the 6tisch ML for the previous
exchanges.)

See below.

On Thu, Apr 19, 2018 at 10:55 AM Christian M. Amsüss <[email protected]>
wrote:

> Hi Mališa,
>
> On Thu, Apr 19, 2018 at 08:24:57AM +0000, Mališa Vučinić wrote:
> > I agree that sending the path exposed at the pledge would be quite clean
> > but I don't get the benefit of it since we need to hardcode /j at the JRC
> > anyways. For the JRC and the pledge to be interoperable, pledge needs to
> be
> > hardcoded with /j to send the Join Request, and the increment for the JRC
> > to be hardcoded with /j to send the rekeying request to the pledge is
> > minimal IMO.
>
> I think it's one thing to mandate a path to the JRC (which is part of
> the "any-cast group" 6tisch.arpa, which by the way is a trick I find
> very neat), and another for the pledge.
>
> What does ease my concerns a bit is that those resources are anyway
> limited to what I'd call the virtual host induced by the OSCORE context.
>
> > > If CBOR is used to glue the two things (the request hex-stuff, I don't
> > > remember what exactly that was) and the path) together, 16 bytes of
> > > hex-stuff and 1 byte of path become 21 bytes of payload.
>
> Side note: Looking through -05 again I noted that the "hex-stuff" I
> mentioned is already in a join_request CBOR; the overhead associated
> with the path would be further decreased.
>

Yes, that's right, we could extend the join_request struct, if we get
pushback on this.


> But yeah, sticking with /j with the option of adding that to
> join_request if there is pushback sounds like a viable option right now.
>

Sounds good!


> > I am considering POST for both the initial Join Request, and for the
> > rekeying request. That should be fine, right?
>
> The requests for join and re-join from the node to the JRC sound good
> with POST. Who would send a re-keying request -- the node or the JRC?
>

Current text specifies that rekeying is triggered by the joined node, but I
am talking about the update to the text where the rekeying request is sent
by the JRC to /j resource exposed by the joined node. POST should be fine
there, as well, right?
_______________________________________________
6tisch mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6tisch

Reply via email to