On 24 Sep 2011, at 02:20, Keith Moore wrote:
To me it seems clear that the risks associated with this proposal are less
than the other risks. Software that assumes that IPv4 space other than RFC
1918 space is unambiguous will break in either case. But at least with this
proposal, there's a well-defined and easily-understood path to fix such
software to minimize the breakage.
+1, but I'm a little bit concerned about transition mechanisms which depend on
exclusively upstream NAT functions rather than in addition to the CPE, i.e.
DS-Lite. As it stands, the two cases can be distinguished, and it seems to me
to be against this proposal, for that situation, since private addresses will
probably provoke "Find my public address" semantics in existing software rather
better than "Public-seeming" addresses on directly-attached hosts behind a
Otherwise, yep, it's inevitable.
Ietf mailing list