>>>>> On Mon, 16 Aug 2004 14:13:14 +0530,
>>>>> "O.L.N.Rao" <[EMAIL PROTECTED]> said:
> I support the argument of optimizing NUD's RA processing.
> However, the RS-RA exchange is not a frequent one to happen. Hence the
> proposed optimization may not be of great advantage. Also, I am
> wondering if it is going to create some IOT issues with old-RFC
> implementations. The simple example of such an IOT issue is, TAHI scripts
> expect a NS to be sent in some cases where TN guesses that NUT is in
> STALE. After that even if RS-RA exchange has happened, TAHI thinks
> the NUT is in STALE and test case proceeds from there. If we introduce
> this optimization, TAHI is likely going to show the test case result as
> FAIL as the new-NUT is not going to send any NS after RS-RA exchange.
> Hence, I would request you to look much deeper into that optimization
> before going to consider it.
> Your comments are highly welcome :).
Basically, I don't have a strong opinion on this. As you mentioned,
this is, even if justified, a very minor optimization (in fact, I
even don't know of a router implementation that supports and does
unicast RAs). So, I'm basically not interested in using the
optimization in our own implementation.
However, I don't oppose to introducing the optimization if the issue
is only related to protocol-conformance tools.
JINMEI, Tatuya
Communication Platform Lab.
Corporate R&D Center, Toshiba Corp.
[EMAIL PROTECTED]
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[EMAIL PROTECTED]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
