Subject: Re: How to use properly ipv6 autoconf over a router interface?
To: None <tech-net@netbsd.org>
From: Martijn van Buul <martijnb@atlas.ipv6.stack.nl>
List: tech-net
Date: 05/04/2006 22:01:45
It occurred to me that Konstantin KABASSANOV wrote in
gmane.os.netbsd.devel.network:
rules.
>
> The length of the prefix advertised can be less than 64 bits. So we can
> imagine other ways to redistribute longer prefixes issued from this =
> one...

1) The advertised prefix *cannot* be more[1] than 64 bits, because of RFC 2462
2) the autoconfig'ed host gets a single address, not a range. There's no way
   to autoconfig a range.

>
> I could agree, but nobody says it is impossible for such protocols to
> appear...=20

Surely that's not impossible. But that wasn't the point. The point was that
autoconfigured routers are considered to be a Bad Thing (TM) and don't make
sense. And within autoconfiguration as we currently have it, that's a very true
statement.

[1] Not less. Less would mean "having a less restrictive netmask".