In your previous mail you wrote: > About your proposal, if we set the globally unique bit on, > you get IIDs > which are no more globally unique. If it is off, this won't avoid > collision... The only advantage is the clarity, i.e. last argument. The state of the bit is not the determining value in uniqueness,
=> hum, the bit is global/local scope and last address architecture draft speaks about uniqueness... so I believe the bit should be set to local like it is for ISATAP. => I agree but in this case collisions won't be avoided by construction. They would still be as unique as any other IANA registered value. => I don't understand your argument... or do you say that they'd still be as unique as the ::1 address for instance? Matt Crawford wrote: > Which would be <prefix>:0200:5eXX:XXYY:YYYY in IPv6 interface id > form. The XXXX portion need not be equal to fffe (but should not be > ffff). I agree it should not be ffff or fffe, but would set the u/l bit off. If they need to exist we should stick them in 0000:5eff:fdyy:yyyy. Again, I have not been convinced there is any value in defining well known host addresses (anycast prefixes yes, hosts no). => this is another argument... I'd like to get the answer from DNS/SLP discovery people (with DNS or SLP all other services can be discovered so well-known addresses are not needed, i.e. the problem stands only for the very first step). Regards [EMAIL PROTECTED] PS: can I say in the HIP mailing list that DNS/SLP should be used in place of a reserved/pre-allocated/well-known set of IIDs? It seems we can reach a consensus about this specific question, i.e. send objections to me or (better) the mailing list ASAP. -------------------------------------------------------------------- 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] --------------------------------------------------------------------
