RE: mech-v2-05pre

["Karen E. Nielsen (AH/TED)" <karen.e.nielsen@ericsson.com>]

Hi Brian, Pekka, All

(Pekka, you're explicitly addressed here in your capacity as mech-v2 editor.)

Now that we are discussing mech-v02 I wonder if I can take the chance
to take up an issue which have bothered me for some time. 

Mech-v2 concerns configured bi-directional IPv6-in-IPv4 tunnels. These
tunnels are point-to-point links in terms of RFC 2461 and mech-v02 talks
about the behaviour of IPv6 ND mechs, NUD in particular, over these tunnels.

Now, 6to4, RFC 3056 in Section 3.1 refers to RFC 2893 for how NUD should be handled
on 6to4 pseudo interfaces.

When we made our implementation of 6to4 a while back we interpreted
this as to refer to what in RFC 2893, Section 3.8,
 was said about unidirectional tunnels, because although 6to4 tunnels
may in some senses be considered bidirectional 
they are not point-to-point links in the terms of RFC 2461, 
as link-local multicast isn't supported.

Now, some people have let me know that this may not be the correct interpretation of what
was intended to be said in Section 3.1 of RFC 3056, and this certainly complies well with the fact that the unidirectional part of RFC 2893 is being deprecated.

But it still makes me wonder what then is the intend of what is being said in section 3.1 
of RFC 3056 and how to handle this reference now that Section 3.8 of mechv02 explicitly
refers to the point-to-point link capacity of the tunnel link - ?

I apologize in advance if this has been discussed at length in the early stages
of the work on mech-v2.

Thanks, Karen