Hi Tony,
The point is that those commenting against SL don't run a real network. There will be filtering done in real networks. This filtering creates addresses and/or prefixes with a local scope of applicability. IE: There will be local scope addresses in any case. The only question is if we have a well-known prefix that everyone filters on, or random values that require explicit n-way coordination. This also affects applications that might want to leverage the prior knowledge of the existence of such a filter.
On the contrary, some of the pushback on site-local addressing has been from people who run real networks. One of the deciding factors in the WG meeting discussion was the statement from a few real network operators that they don't need a special prefix for non-connected networks -- since they'll have to renumber when they connect, anyway, they could just use a random prefix on their disconnected networks.
There is only one reason why using a random prefix on disconnected IPv4 networks causes problems when those networks become globally connected -- NAT. Folks do not choose to renumber their networks, they use NAT instead. And, if they've chosen a random prefix, they won't be able to connect to outside networks that use that same prefix.
We hope to avoid this problem in IPv6, by avoiding the use of NAT.
If you accept that you will have to renumber when you move from a disconnected state to a globally connected state, why does it matter whether you used a well-known prefix or a random prefix before the renumbering?
Margaret
-------------------------------------------------------------------- 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] --------------------------------------------------------------------
