Hi Jinmei, 

Sorry about the confusion.

----- Original Message -----
From: JINMEI Tatuya / ???? <[EMAIL PROTECTED]>
Date: Thursday, August 5, 2004 7:31 am
Subject: regarding some comments on the M&O draft

> Hello,
> 
> I'm not sure if I understand your comments on
> draft-daniel-ipv6-ra-mo-flags-00.txt in the wg meeting.  (I've checked
> the jabber log to be sure, but I'm still not 100% sure).  Would you
> mind to repeat those?

I was worried about the SAA M and O flags
having information available such that 
M could be used to invoke stateless configuration
procedures on a host (since 2462's statement
in 5.5.3 about getting addressing and other information).

I talked to Dave Thaler about this after
the session and (I think agreed) that if 
the M flag is only used for indicating
that stateful address configuration is
available,  and the O is always set when 
non address configuration information is available,
then the names are OK, and we only need to look
each flag individually.

I was concerned that M|O could be used to 
invoke DHCP information-requests
(rather than just O).

If we only mean 'managed addressing' with M, and
Other config with O  then each flag is aligned
with a particular flag.

If we have flags which can be used for both
purposes (like if M=1,O=0 allows hosts to send
information-requests) then our policy names need
to be adjusted.

 
> To provide some answers at the moment:
> 
> As for the comment on policy 1 (always try DHCPv6/stateless-DHCPv6),
> it's true that the policy might cause weird effects (but I'm not sure
> if it's specific to mobile nodes).  So I agree that we should be
> careful when taking this policy.  Also, we may even want to avoid
> introducing the policy from the beginning.  (Please note that this is
> the first attempt to provide the missing piece of the details usage
> about the M/O flags as a result of the rfc2462bis discussion.  We'll
> definitely need to discuss many technical details on the first
> proposal.)

Removing policy 1 may be an option, but it's
may also be good to have a way to provide an
override capability if needed.

I think that hosts SHOULD NOT use policy 1 by
default though.  If they do, it should be by
explicit configuration.

> Regarding your second point, I'm even not sure about the point...from
> the jabber log:
> 
> [18:28:54] <timchown> daley: m and o both indicate stateless info 
> is available
> [18:29:52] <timchown> daley: o and m policy names my be bad names. 
> goals of original flags was different
> [18:30:00] <timchown> (my = may)

The 'my' is just how I pronounce 'may'...
 
> If that's just a naming issue, that's fine.  I'm willing to rename
> them if someone can offer better ones.   But I don't understand the
> comment on the "original goals".  I'm even not sure what the "goals of
> original flags" means, but, in my understanding, we are going to
> describe the detailed and desired usage of the M/O flags as a BCP,
> considering the latest standardization/deployment status.  In that
> sense, it's not surprising that the result is different from the
> "goals of original flags".  The important point is, IMO, that the
> result makes sense and meets today's deployment scenario.

I agree.

> (Of course, since this is a (candidate of) derivative work from
> rfc2462bis, we'll need to care about the "original intentions")
> 
> Thanks for your clarification in advance,

I hope that there has been some clarifcation.

Greg


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

Reply via email to