On 26.6.2015 18.41, Juliusz Chroboczek wrote:
Markus,
I still don't understand the intent of the "ad hoc" interface type.
If the ad-hoc interface is designed for non-transitive links, then the
draft should say so (in which case I'll be glad to provide you with
suitable prose). However, in this case it should also identify the bits
of DNCP which are not applicable to non-transitive interfaces (Section
6.1.5, obviously, but I don't understand DNCP well enough to say if
there are others).
Clarifications on the text are obviously welcome and even needed :)
Due to renumbering of sections, this is bit awkward, but _as far as I
know_, the only part that requires transitive interfaces _in DNCP_ is
the support for dense broadcast links.
6.1.5 in dncp-06 is 'neighbor removal', and it makes no transitivity
assumption. For each individual node, an endpoint is a local construct
(e.g. 'this meshy eth0 interface I want to multicast+unicast on') and
the peers that are present on it (nodes N_1.. N_n that happen to have
been recently reachable / we have received keep-alives from / we know to
be there by other means).
Section 4 speaks of "ad-hoc mode" -- does that mean 802.11 IBSS? If it
is, then you should use the proper terminology, but I think that support
for non-transitive links should not be restricted to 802.11 (even if the
implementation is limited to it).
Even implementation isn't limited to it.
And sorry if I sound like a broken record, but I would like the ability
to set up a router-router link with less than a full /64 allocated to
it, at least in the ad-hoc case.
I am not sure about the draft text about what it should say, as non-/64
seems to be a political hot potato (there are drafts on this too), but
at least the implementation isn't limited to this (given configuration)
and the draft allows for it as well (given configuration). The default
of /64 seems still sane to me though, given no configuration.
Cheers,
-Markus
_______________________________________________
homenet mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/homenet
