Date: Tue, 22 Jul 2003 18:02:03 +0200
From: Francis Dupont <[EMAIL PROTECTED]>
Message-ID: <[EMAIL PROTECTED]>
| => this is an old issue.
Yes.
| IMHO the best solution is to consider the MAC-48 -> EUI-48
Is there such a thing? I thought that MAC-48 and EUI-48 were
distinct (though very similar looking) number spaces.
| BTW the EUI-64 stuff is a big joke:
| *no* IEEE layer 2 uses EUI-64s as addresses today
It certainly is currently. There isn't likely to be a link layer
that uses EUI-64 until something remarkably new appears (something
that no-one wants to be compatible with ethernet in any way) but
which is designed to be of similar utility to ethernet. This
combination is perhaps unlikely...
But it does no real harm to allow for EUI-64 - we want IPv6 addressing
to last for MANY years into the future - EUI-64 is certainly a plausible
addressing structure for some future link layer to adopt (given that
it is defined, etc). If it never happens, big deal (we're not about
to switch IPv6 addresses to be 112 bits, and nor do we need more than
64 bit prefixes (48 bit site identifiers)).
On the original point - the stupid thing is that Deering (I think) had
the transformation defined originally using FFFF (the correct value to
convert MAC-48 to EUI-64), but the WG was then told that was the wrong
value, and FFFE was the one that should be used. And we believed it...
I'd suggest now just documenting the thing, perhaps reserving FFFE for
EUI-48 to EUI-64 conversion in IPv6 addresses (should this ever be
needed). IPv6 addresses don't use "true" EUI-64 anyway, they have
the U/L bit inverted. If it is OK to invert one bit, it must also
be OK to (sometimes) invert another bit... It is weird, but it
costs nothing beyond the explanation in the doc.
kre
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
