Folks,

It was a pleasure to meet Erik face to face and chat - I also appreciate
his time. 

We have already agreed about 4 months back with Jari that we are not
suggesting any tweaking of text in RFC 4861 via our 3rd draft 
(http://www.ietf.org/internet-drafts/draft-wbeebee-nd-updates-00.txt) -
my slides also said, we are not actively pursuing any updates to be made
to RFC 4861 - I also agree with Erik the updates would get lost in the
sheer volume of that RFC; our updates draft should not be even the
subject of discussion for 6man.

We only wanted folks to especially read the first draft
(http://www.ietf.org/internet-drafts/draft-wbeebee-on-link-and-off-link-
determination-00.txt) that makes interesting reading in just the items
1-7 of section 2. Some items in the list highlight IPv4 centric mind-set
of implementations. Right now we just have very explicit text in this
on-and-off-link draft for IPv6 mistakes. Erik's suggestion is well-taken
that we could also explain the differences with IPv4 in these cases we
discuss.  

On another note, the example that Fred Baker gave in his IPv6 SAVI BOF
for spoofing gotchas is listed as an example scenario in item 5 of
section 3 of our on-and-off-link draft mentioned in the URL above. We
presented this example to 2462bis-8 as a scenario to show why skipping
DAD did not make sense. 

I and Wes will work with Erik to make some high-level explanatory text
available. 

Thanks very much Erik. Also thanks to all who replied on the thread.

Hemant

-----Original Message-----
From: Erik Nordmark [mailto:[EMAIL PROTECTED] 
Sent: Wednesday, December 05, 2007 6:37 PM
To: [email protected]
Cc: Hemant Singh (shemant); Suresh Krishnan
Subject: Re: Off-link and on-link


Hemant and I met this afternoon and he explained the details of what
they've observed, which helped me a lot more than the discussion in the
WG meeting.

Based on this I think I understand the root cause of why some
implementors get some things wrong.

The IPv6 subnet model is quite different than IPv4 in that it is
optional to have any subnet prefix. In IPv4 there is always a subnet
prefix associated an IP address (at least on anything but a
point-to-point link).

While one can *infer* this from all of the text in RFC 4861, it might
make sense to have an RFC which
  - makes this difference very explicit
  - gives examples of some implementation mistakes we've seen as a
result

I don't think tweaking the text in RFC 4861 will necessarily help, since
it gets lost in the volume of that RFC.

I told Hemant I'll help to make sure that high-level explanatory text is
included.

    Erik

--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to