Le 15/04/2010 05:18, Philip Levis a écrit :
On Apr 14, 2010, at 3:32 PM, Alexandru Petrescu wrote:
Le 14/04/2010 22:23, Philip Levis a écrit :
On Apr 14, 2010, at 10:21 AM, Alexandru Petrescu wrote:
Hi, I came across this draft coming from the ROLL space where a
proposal exists to use the Flow Label changed enroute maybe.
Alex,
I think you're misrepresenting ROLL here, even though (after your
objections there) the use of the flow label was very clearly
described a few days ago. Changing the flow label enroute is not
on the table in ROLL; it was a while ago as a crazy idea, then
everyone realized the issues it would create.
Sorry for misunderstanding...
(only a day ago it was stated in ROLL list that changing the flow
label enroute is possible as long as it is restored...)
Sure, it's possible, that doesn't mean RPL is going to do it. Doing
so requires having somewhere in the packet to store the original
value, which necessitates a header; if there's a header, there's no
point in changing the flow label. It's my understanding there are
some proposals out there for doing this, but in other domains --
this was mentioned at the mike in Hiroshima -- but that by no means
indicates RPL will.
Ok, this is a discussion about intentions. I am not aware of the
discussions in Hiroshima and intentions of future text, thanks for
having clarified this.
The current text in the RPL 07 draft says bit O oh of Flow Label set to
0 zero by Host and reset and set again by several intermediary routers.
I guess this text will change.
Alex
Phil
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
