> A DHCPv4 extension would work, somethink like marc blanchet's TSP
> protocol would work also. Maybe the first think to do is to work on some
> requirements for such a protocol.
Indeed, we have to get a better picture on what are the problems and
non-problems.
For example, DHCP extension works only in scenarios where you obtain the
tunnel connectivity from your direct upstream ISP, the same one that
provides IPv4 connectivity.
In many cases, this is a non-problem: it is often the case that such
IPv6-active ISP's could provide v6 access using different means already --
no all though -- if the IPv6-conscious ISP just has to keep natting
routers as CPE's and can't change them in the short term, this could still
be valuable.
So, we need to have a picture on the scenarios we're trying to address.