Date:        Fri, 04 May 2001 19:27:03 +0900
    From:        [EMAIL PROTECTED]
    Message-ID:  <[EMAIL PROTECTED]>

  |     we do not implement "partly host, partly router" behavior.

That I can understand, it seems like a pretty weird thing to want to
me (though conceptually possible, and perhaps even an interesting case
to consider in protocol design, not a beast that many people are likely
to actually want to use).

  |     also, since RFC2462 basically assumes single-interface host,

Huh?

I just re-read the thing, and I can't find any assumption like that at
all.

In fact, the first sentence (after boilerplate, abstract, etc, is...)

   This document specifies the steps a host takes in deciding how to
   autoconfigure its interfaces in IP version 6.

"interfaces", in the plural.   Sure, there are a few places where the wording
is perhaps a little loose when view with that possible assumption in mind
(which I had never done before I'll admit), eg:

   If a node determines that its tentative link-local address is not
   unique,

(3rd para, section 4) - which probably really should say

   If a node determines that the tentative link-local address for one
   of its interfaces is not unique,

and

   To speed the autoconfiguration process, a host may generate its
   link-local address (and verify its uniqueness) in parallel with
   waiting for a Router Advertisement.

(last para of section 4, just before section 4.1), which would probably
be better written:

   To speed the autoconfiguration process, a host may generate the
   link- local addresses for its interfaces (and verify their uniqueness)
   in parallel with waiting for a Router Advertisment.

but everywhere substantive in the spec I could find it always talks of
interfaces, and specifies things to be done per interface.

eg:

5.  PROTOCOL SPECIFICATION

   Autoconfiguration is performed on a per-interface basis on
   multicast-capable interfaces.  For multihomed hosts,
   autoconfiguration is performed independently on each interface.

which is about as clear as it can be I think that multi-interface
hosts are expected to be supported.

What leads you to the "assumes a single interface" conclusion?
Whatever that is, certainly needs to be fixed.

  |     we do not really care about the following cases:
  |     - autoconfigured host with multiple interfaces

You might not care about them, but they seem to work for me?   That is,
I run KAME, I have nodes set to "autohost" with more than one interface...

  |     - chimera (part-host, part-router) node

Understandable.

  |     - autoconfigured router

that would be broken (aside from link local autoconfig, and I suspect
that works in KAME too).

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]
--------------------------------------------------------------------

Reply via email to