Hi, I have a tunnel to my box at home, and a subnet routed to it for all the remaining boxen. For consistency, I was wondering whether it would make sense to use the address from the internal, "subnet" interface on the tunnel endpoint only, as the status quo is that the IPv6 stack selects the one that has the longest common prefix with the destination, so connections from my host seem to originate on two different addresses. Also, I'm not sure it is necessary to use globally routed adresses on the tunnel interface if the first thing one does is add a subnet over it. So, would it make sense to "force" the address scope down to "link" on the tunnel (if that is at all possible; the Linux IPv6 stack seems to ignore me anyway)? Would it make sense to have an option to "give back" the addresses on the tunnel link, as they are unneeded as soon as a subnet is present, in order to conserve address space :-) Simon