RE: dual stack & IPv6 on by default

[Mika Liljeberg <mika.liljeberg@welho.com>]

Hi Jim,

[The following quote concerns RFC3484 and destination address selection]

> >An implementation can't really know that a destination address 
> >is unreachable, unless it does a route search and, failing to 
> >find one, proceeds to try to ND all routerless links as well. 
> >Is that what we should do?
> 
> I think today this is the bottom line only answer.

That's too bad, because it means that the DNS resolver would have do
this for the whole candidate destination address list. That takes time.
I would much prefer do a route search only.

> But your mail thread makes me wonder if we need to do more and I think
> so.

RFC1122 clarified and corrected several small details in the IPv4
specifications. I think the node requirements spec could do the same for
any fuzzy bits in the IPv6 specificatons.

> But it would have to be in the context of ND.  ND is a requirement to
> use IPv6.

I agree. This is just a detail that needs to be clarified somehow.

> Feels like an edge that could be "do this if ND don't work"?

No. It's a very specific case of "how to implement the following bit of
next-hop determination" in a host with multiple network interfaces and
how it relates to RFC3484 and destination address selection:

	If the Default Router List is empty,
	the sender assumes that the destination is on-link.

Until there is a clear understanding, we are sticking with:

	If the route search fails,
	the sender assumes the destination is unreachable.

I haven't seen comments from any other implementors. It would be very
interesting to hear if you have solved this and how you implemented it.

Thanks,

	MikaL