Title: RE: rfc2553bis comments

Francis Dupont writes:
> => I don't believe we should specify two different interpretations
> but obviously we have to do the choice between:
>  - use interface indexes and make scope IDs not very useful for the
>    receiving side
>  - use traditional (ie as for unicast) scope IDs and make the sin6_scope_id
>    field nearly useless for emission to destinations to a scope larger
>    than link-local (the only thing I can see is to check interfaces
>    specified by IPV6_MULTICAST_IF are in the right zone)
> I was in favour of the first solution, now I believe the second is better.

It's not useless on emission, since it's not required to do IPV6_MULTICAST_IF.
If you don't do IPV6_MULTICAST_IF, the meaning would
be to send via some interface in the set identified by scope_id.

However, when IPV6_MULTICAST_IF is done, the most you can use it for
is a sanity check and fail the send if you try to send out an interace
not in the given scope.

-Dave

Reply via email to