Re: [Autoconf] WC consensus call for RFC5889 modifications(Fwd:Forgotone [Was: RFC 5889)

["Charles E. Perkins" <[e-mail]>]

Hello Emmanuel,

To answer your implied request:

On 8/9/2010 10:29 AM, Emmanuel Baccelli wrote:

>     (c) the [autoconf] address allocation protocol design
>         should proceed on the assumption that non-routing
>         hosts are out of scope.
>
>
>
> here on the other hand, I think I disagree.  ...

> So at this point, either (i) there is priori art that
> provides autoconfiguration for nodes which have heterogeneous
> capabilities, and then let's consider it, ...

> I remember you mentioning that there is indeed prior art.

Here is a link to prior art, which I mentioned
in a previous email to you on this list:

>  .......         You might check [from year 2001]:
>     www.cs.ucsb.edu/~ebelding/txt/autoconf.txt

What do you think about that?

But, maybe more to the point, I know I have seen a
good half-dozen distinct approaches to this problem.
They have various advantages and disadvantages, but
they share one trait:
- they all provide addresses for both forwarding
   nodes and non-forwarding nodes.

Right now I can't think of a good approach that
actually does disqualify hosts.

Of course it's not a democracy, and the other authors
are not in the IETF.  I guess they don't like to have
five-year debates about what is a router and what is
a link.  But they are very sharp people.  I would have
thought that their insight, experience, development,
and testing might count for _something_.

Regards,
Charlie P.