Title: FW: rfc2553bis comments

Looks like the last try went out as HTML.
Resending, sorry for duplicate.
-Dave

-----Original Message-----
From: Dave Thaler
Sent: Tuesday, June 27, 2000 9:53 AM
To: Francis Dupont; Steve Deering
Cc: [EMAIL PROTECTED]
Subject: RE: rfc2553bis comments

Steve Deering writes:
>      if the local indexes for the interfaces plus all zones
> (i.e., links,
>      sites, and any additional multicast zones configured on
> the node) are
>      given distinct values out of the same scope_id space,
> and if one index
>      value (say, 0) is reserved to mean "unspecified".  Then,
> I think we could
>      probably get rid of IPV6_MULTICAST_IF altogether (i.e.,
> deprecate it).
I really don't like the idea of sharing the same numbering space
between interface ids and scope ids of higher levels, and requiring
uniqueness among them.
My reasoning is that ifindices are a link-layer id (per their original
definition in MIB-II), whereas site ids are network-layer (and IPv6 only)
ids with a very different meaning.  It's a layer violation to ask for
uniqueness among them.
If the implementation used interface ids that were not ifindices, that would
be a different thing, but then if_nametoindex() and related functions wouldn't
be helpful for an app wanting to choosing a link.
Francis Dupont, responding, writes:
> => the only reason we should have to keep IPV6_MULTICAST_IF is an
> interface specification can be finer than a link one...
I agree with that reason, but obviously I don't think it's the
only reason :)
>      On reception, the upper-layer should be told the arrival
> interface,
> => how? The sin6_scope_id should contains the index of the zone of the
> address scope (this rule implies global addresses in any
> cases will get
> the zero sin6_scope_id in recvfrom() because there is only
> one global zone).
I believe the reception interface (for both unicast
and multicast) can already be obtained if you use recvmsg.
 
>      from which it can also determine the arrival zones of
> each possible
>      scope, since each interface can belong to only one zone
> of each scope.
-Dave

Reply via email to