On Thu, Apr 03, 2003 at 10:08:57AM +1000, Andrew White wrote: > > Is anyone interested in pursuing this design? > > Well, I have an implementation. > > If Bob is happy, I'd like to grab most of his text (since it's better > written than mine) and wrap it around my bit-ordering proposal. > > > > - If the /16 is well known, it can be plugged as "least preferred" in > > the address selection rules. > > I still have qualms with this. I'm of the opinion that if site-locals exist > in a network then hosts should prefer them (iff both hosts have them); the > assumption is that SL addresses are at least as stable as externally sourced > addresses.
Just a side note--my main point is below, but why is this the case? If a router is still advertising a global prefix even though there is no global connectivity, two local hosts can still communicate using that global prefix, correct? > Of course, this policy will potentially be inappropriate for applications > that do address forwarding across a site boundary, but I'd argue that these > are the special case. Apps that do address forwarding NOT across a site > boundary will be happy with the standard policy. I'm sure it's obvious, but apps that do address forwarding across site boundaries should probably just be aware that they are P2P apps and thus use a socket option that makes them prefer global addresses. If this were a part of the API documentation, I think this would be fine. Best Regards, -jj -- Hacker is to software engineer as Climbing Mt. Everest is to building a Denny's there. -------------------------------------------------------------------- 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] --------------------------------------------------------------------
