>NAT should be a deterministic mapping in the spatial dimension,
>with only long-term changes to that mapping.

You don't have enough port numbers for this in a many-to-one situation.
There must be a termporal component.  For TCP it works OK, for 
UDP it doesn't work.  TCP has a well defined start and stop.  UDP
doesn't.

>This is likewise a problem without any NAT at all.   If one is
>assigned addresses dynamically, and other parties don't have
>knowledge that a change has happened they cannot know that they
>should re-query the DNS.

Yes, another evil propigated due to a limited IPv4 address space.
If a machine doesn't change location, why should it change IP?

Oh, and rpc2 just treats the new IP as a new host and things work
relatively well.  I have moved a coda client between several IP
without restarting coda and haven't had any problems.  I'm not
going to say that problems wouldn't come up, but under NAT I can't
get anything going.

>Yeah, if you have a PAT rule that compresses everything into one address.

If one doesn't have some sort of compression, why use NAT?  If one has the
same number of IP address comming out of NAT as is behind it, why have it???
The reason I'm actually using NAT is because each IP costs $10/month
above my basic 1 IP connection.  I can't afford on IP per machine.  Therefore,
I must have compression.

>Eh, why doesn't it respond to an embedded DNS name?

There are no embedded DNS names or IPs.  It gets all the IP information
from the original UDP packet.

>Can another protocol than rpc2 be substituted, or is there
>no real way to swap one data-carrying protocol for another?

The only possible "easy" way is to do rpc2 protocol over a tcp connection.
And the biggest reason to consider doing this is because of NAT.

Finally, the performance of coda would suffer greatly if it had to
add DNS queries in other than for initial contacts.

And consider this my last e-mail on this subject.  I consider NAT
evil and you are not likely to change my opinion.  Isn't there a
saying like "A man convinced against his will is of the same
opinion still."  I guess this goes both ways.  :)

-- 
Phil Nelson                       NetBSD: http://www.netbsd.org
e-mail: phil@cs.wwu.edu           Coda: http://www.coda.cs.cmu.edu
http://cs.wwu.edu/faculty/nelson