----- Original Message ----- From: "Robert Elz" <[EMAIL PROTECTED]> > > If I didn't deprecate the old address, then I'd have two preferred addresses, > and connections started at my node would be just as likely to use A1 as A2, > and that isn't what I want - since the address is changing, everything > should use A2 in preference to A1, when that is possible. >
This is a rather narrow-minded view, that makes many assumptions, which do not necessarily hold, especially with 128-bit DNS and addresses. As an example, if the addresses contain **version** information and/or **time-stamps** the end-to-end agreement can be that one version of the software will use the addresses that match it until some time is reached and they roll over to the next version. With the "toy" Internet (i.e. 32-bit) addresses, there was not a lot of room to store extra information. As an example, software can look at 2002 addresses and prefer those over 2001 addresses. Next year, the software may prefer 2003, and deprecate 2001 and 2002. Jim Fleming 2002:[IPv4]:000X:03DB:...IPv8 is closer than you think... http://www.ican.org/what's_new!!!.htm http://www.iana.org/assignments/ipv4-address-space http://www.ntia.doc.gov/ntiahome/domainname/130dftmail/unir.txt -------------------------------------------------------------------- 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] --------------------------------------------------------------------
