Date: Thu, 27 Jun 2002 15:12:21 -0700
From: "Tony Hain" <[EMAIL PROTECTED]>
Message-ID: <[EMAIL PROTECTED]>
| I agree no address should ever be ambiguous, but that doesn't require
| global uniqueness.
This all depends upon the frame of reference for "ambiouous". You're
agreeing that addresses should be unambiguous from a particular point
of reference. Keith wants addresses to be unambiguous from any point
of reference - in particular, so an app can send an address from one
node to another, without having to worry about whether it will mean the
same thing when it arrives as it dopes when it was sent.
My take on that is that limited scope addresses simply shouldn't be used
that way - it may be true that using DNS names as a reference to ship
around isn't good enough (personally, I'm not convinced that if it
ever became important they wouldn't be good enough - most of the reported
problems are due to laziness or poor knowledge, which affect anything
when there's no incentive to do better).
But I see no reason not to ship around only known global addresses
(ones expected to be globally reachable), and then allow apps that desire
to use more limited scope addresses to discover the limited scope address
from the global.
Incidentally - this also relates to your comment's on the icmp-name-lookups
draft ...
It
also doesn't make any sense to use a mechansim that is clearly getting
packets to the dst node to ask it for other addresses that might be used
to get packets to it. THe only possible use I can see would be to allow
shifting to a different pair for route selection reasons,
Another possible use is to shift to a different pair, of a different scope
(not caring about the routing used in the slightest). That is, you ask
the global addr to return you the site local addresses.
As long as it is obvious on the face of it what is a global address,
and what is limited scope (which matching fe80::/9 tells us now), having
apps only tell others about global addresses, and never passing off a SL
addr to another app looks real easy to me.
The only requirement is that a global address actually exists to use.
kre
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
