Date: Tue, 13 Aug 2002 13:08:17 -0400
From: Margaret Wasserman <[EMAIL PROTECTED]>
Message-ID: <[EMAIL PROTECTED]>
| I am very familiar with two implementations that generate IIDs from EUI-64
| identifiers and set the 'u' bit accordingly, and I am sure that most
| other implementations do this as well.
|
| That's the only think that the addressing architecture specifies that
| you can/should do with the 'u' bit...
That's fine, and I'm not advocating removing that part of the draft (nor
were you when you posted your original message, I didn't think).
What I want to remove are the words that say that an address with the 'u'
bit set contains an IID that has global scope (which isn't actually defined
anyway, that is what does "has global scope" actually mean?)
That is, I have no doubt the implementations set the 'u' bit the way
you say they do, so do the ones I have seen.
But do they also clear the 'u' bit when the IID doesn't come from an EUI-64?
None I have seen do that. That is, do as I did (and showed) in the message
I sent yesterday, try configuring
prefix::0200:0000:0000:0001
(or anything you like in the bottom 48-56 bits) and see if the
implementations you are familiar with either reject that (give an error)
or silently clear the 'u' bit.
If they don't - they're not implementing 'u' the way the arch doc claims
it should be implemented.
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]
--------------------------------------------------------------------
