>
> > From: JINMEI Tatuya / =?ISO-2022-JP?B?GyRCP0BMQEMjOkgbKEI=?=
>
> > conditions, just delaying Neighbor Solicitation would cause
> > congestion by the MLD report messages. The congestion would then
> > prevent MLD-snooping switches from working correctly, and, as a
> > result, prevent Duplicate Address Detection from working.
>
> Just a refresh of my opinion on this (so that people don't forget):
>
> Sending MLD reports on link local multicast groups is silly, and
> SHOULD NOT be required.
>
> It is not nice to require modifications and warts to perfectly working
> RFC due to layer breaking entities like MLD-snooping
> switches. Especially, when exactly the same information would be
> available to them from the DAD NS. Join for "solicited node multicast
> group" has exactly the same information as DAD NS.
i understand your logic, but disagree. what you are saying is
"we can make MLD-and-NS-snooping switch and require no MLD for NS".
MLD-and-NS-snooping switch is also a layer breaking entity.
itojun
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[EMAIL PROTECTED]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
