"Michel Py" <[EMAIL PROTECTED]> writes: > >> Michel Py wrote: > >> IPv6 over ethernet: stick to exactly /64. Probably for TR and > >> FDDI too. > >> IPv6 over foo: it might be desirable to get a lower value > >> (make it fit on a nibble or byte boundary) _if_ accompanied > >> by RFC2373 modifications or new text that define a fixed > >> aggregation boundary for routing purposes. > > > Alexandru Petrescu wrote: > > I like these, FWIW. This would imply changes in the meanings > > of Prefix 001 and Prefix 000. I doubt the intended meaning > > can be to have Ethernet under Prefix 001 and all other L2's > > under Prefix 000. > > I think that it is all under 001. Allocating another 1/8th of > the v6 address space (000) to save pennies on the 001 does not > make a lot of sense to me.
Right, that reservation is not reasonable. My remark with the 000 prefix relies on the fact that this prefix is the only one today loosening the 64 bit Interface ID length: "Global unicast addresses that start with binary 000 have no such constraint on the size or structure of the interface ID field." This is made explicit in several places in the draft and I was guessing that it is so in order to allow for other Interface ID's than those derived from Ethernet (and not only to accomodate v4 addresses) under prefix 000. Probably it makes more sense to loosen all constraints on the size or structure of the Interface ID field of a 001 Glocal Unicast Address other that it should not exceed 64 bits in length. And leave the 000 addresses for v4 only. Alex -------------------------------------------------------------------- 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] --------------------------------------------------------------------