Steve Deering replied:
> At 4:42 PM -0700 6/29/00, Dave Thaler wrote:
> >For example scop 4, being smaller than site-local scope,
> >should use the site-local scope ids. If you're
> >in multiple sites, you can't assume all interfaces
> >are in the same scop 4 zone.
>
> Hmm, yes, I can see how you can use that approach to avoid creating
> zone indices for scop 4 on site-boundary nodes. I was working from
> the assumption that you *would* have scop 4 zone ids on such a node,
> i.e., *if* scop 4 were configured to be used within a site, it
> would have to be configured on all site-boundary nodes.
Re "if scop 4 were configured to be used within a site"...
So if we have a host with interfaces in two sites, and
there's no manual configuration related to scop 4, and
if an app tried to send to or join a multicast group
in scop 4 are you saying that the send/join would fail?
I'd rather be liberal in what we accept here and have it
succeed, which is what I was trying to allow.
> However, you still need the capability to explicitly configure scop 4
> zone ids on site-boundary nodes, in case the node is attached (via
> different interfaces) to more than one scop 4 zone within the
> same site.
Agree.
> And you also need the capability to configure scop 4 zone ids on nodes
> that are internal to a site, i.e., not site-boundary nodes.
I didn't follow this. I was suggesting that you don't need
separate scop 4 zone ids in this config, you use the same ids
as for scop 5 (and you use a disjoint numbering space as Erik
suggested to disambiguate this).
> >So my proposed correction to Steve's rule is:
> >"Whenever dealing with an address (multicast or unicast) of
> >a scope for which you are not a boundary, the appropriate zone
> >is unambiguous, so the sin6_scope_id is the same zone id of
> >the next larger scope for which you are a boundary (e.g., zero
> >if you are not a boundary for any larger scope level)."
>
> That's still not quite right, because if you are a boundary for
> scope x, you are *necessarily* a boundary for all configured scopes
> less than x, since a smaller zone cannot span across boundaries
> of a larger zone.
I agree the wording "are not a boundary" is problematic here.
-Dave
