IPv6 on ATM

[bound@zk3.dec.com]

Dimitry,

>This was not my reading of the "community consensus".  As I remember it,
>the consensus was that Greenville and Peter should attempt to combine
>their proposal and then community will decide between remaining two.
>For one, both Greenville's and Peter's proposals are ATM centric and it
>would be a win if we define solution which works over all NBMA technologis
>as the Atkinson et al proposal does.

No not quite what I heard and we can now go back and ask the chair and
lots of people in the room at Montreal.  The minutes were just a
snapshot of about 1.25 hours and room full of consensus on the direction
to be headed. 

The issue is that NHRP would cause ND and Autoconf to treat the ATM or
NBMA link different than we treat Ethernet or FDDI as examples.  This
means that we all have to have extra or different code at the Internet
Layer to do ND for NBMA.  Peter has proposed a means that will not cause
this result, and will work with Greenville to adapt his proposal to MARs
and other parts in that draft.  

The bottom line is that the community did not want to tell grad students
10 years from now ND and Autoconf works this way on Ethernet and FDDI
and this way on NBMA.  And lots of us host vendors don't want to write
the ) code two different ways either.

NHRP lost a major battle at the Montreal meeting and in the community on
the host (I am not speaking of the router).

Denying that will only procrastinate a solution.  The Atkinson proposal
was deemed unacceptable was my reading and lots of other peoples for the
link case.  I heard the opposite from you that Schulter/Armitage will
resolve that problem with a proposal to the ION group at San Jose.
I know Peter and Greenville are working on this now no one from NHRP has
attempted to work with them do determine where NHRP fits on the router.
For the host we simply don't need it.  

/jim