Re: [Autoconf] Conclusion: draft-ietf-autoconf-adhoc-addr-model-02.txt

[Henning Rogge <[e-mail]>]

Am Mittwoch 17 Februar 2010 19:23:48 schrieb Charles E. Perkins:
> >  Multicast is what typical autoconfiguration
> > 
> > protocols use today without which they'd never work.
> 
> This isn't true.  In every case I am aware of, we could
> have designed for pure broadcast operation.
> 
> But of course you might just say that broadcast is a
> special case of multicast.  In that case, your
> argument completely loses its apparent force anyway.
> 
> > Multicast is what IPv6 got builtin precisely for the reason of
> > autoconfing.
> 
> Nothing wrong with using the tools available to
> perform the function for which they were built
> in the systems where they can be expected to work.
I think most "autoconf protocols" just use a linklocal multicast, which is the 
same as a broadcast without retransmission.
 
> > This draft being silent about multicast spells it's not autoconf, IMHO.
> 
> Well, I strongly disagree, especially since your
> arguments as I understand them are fatally flawed.
> 
> However, I'd be happy to have a non-tree-based
> local multicast group consisting of all members
> within range of a node.  There's absolutely zero
> multicast protocol required to maintain membership
> in this group, conveniently.
Do current IP stacks even have something like "linklocal membership" ?

Henning Rogge