From: Keith Moore (moore_at_no.spam)
Date: Thu Mar 21 2002 - 07:22:15 PST
> > the only new idea I've seen here
> > is to use A records as a guess of the v6 address of a host.
>
> It's a specification, not a guess. It's analogous to the substitution of
> A for MX->A when there are no MX records.
no, it's a guess (even if you've written a specification for how to
do the guess). it's a guess because you don't know whether the host
with that A record actually supports that v6 address - that host may
well have a different v6 address, or not support v6 at all.
> It has to be globally defined, so that a native IPv6 client following
> the specification can reach a server following the specification.
clearly having a standard guess / recommended practice for a v4 host's
v6 address works better than asking the client to make a random guess
about its destination's v6 address.
> This, too, is analogous to the MX situation.
disagree. if a domain has an MX record it's because a DNS administrator
explicitly placed it there. making assumptions about the practices of
a remote host in the absence of explicit information is always somewhat
risky.
Keith
This archive was generated by hypermail 2.1.7 : Fri Oct 06 2006 - 00:00:32 PDT