Re: draft-ietf-ipv6-unicast-aggr-v2-02.txt

[Randy Bush <randy@psg.com>]

> I included some comments on your comment in my original message,
> and Harald responded:
> 
> >>The whole point of this document is to move the current TLA/NLA
> >>assignment rules to historic, removing the fixed internal
> >>hierarchy from IPv6 addresses... It is no longer reasonable
> >>for implementations to assume that there is any hierarchy to
> >>these addresses, other than the bitwise hierarchy inherent in
> >>CIDR-based/longest-match forwarding algorithms. So, I think it
> >>would be misleading and potentially confusing to add hierarchy
> >>fields to the address diagrams.
> >>Of course, in reality, registries and enterprises can and often
> >>do allocate addresses hierarchically. But, that doesn't affect
> >>the structure of IPv6 addresses.
> >
> >My reading of the comment is that the reviewer wants the same thing - to 
> >say that the one who gets the allocation to play with can put in hierarchy 
> >at any bit boundary he/she/it wants, and that CIDR-based longest-match 
> >will make the Right Thing happen. Randy - is that what you wanted?
> 
> Is Harald right about what you wanted?
> 
> If so, would it be sufficient to place a comment in the text that says
> that the prefix and subnet fields may be hierarchically assigned by ISPs
> or site administrators?

yes

> I wouldn't want us to clutter up the diagrams with arbitrary levels of
> hierarchy for each field, as implementations are not supposed to be
> aware of any boundaries within these fields.

what 'fields'?  fp, ok.  beyond that, i think that's the point.

randy