Steve,
> >Perhaps we have two choices:
> >
> >1. treat :: as global
> >2. does (explicitly) not define the scope type (level) of ::
> >
> >Even the choice 2 will make the document clear, and it will not cause
> >a problem in a practical point of view.
>
> I think of :: as indicating the absence of an address, in which case
> it doesn't have any scope. For those who feel the need to assign a
> scope value to ::, it's probably safest to treat it as link-local
> (for example, to ensure that a router receiving a packet with a
> source address of :: is not forwarded to another link).
>
In thinking about how these facilities are going to be used by applications,
it seems that we need a way to express a desire to accept connections or
datagrams from any address in a specific zone. We also need a way to express a
desire to accept connections or datagrams for a specific address within
any zone.
That is, I would like to be able bind a socket to any of:
::%<specific type>.<specific zone>
::%<specific type>.<any zone>
::%<any type>.<any zone>
<specific address>%<specific zone> (The address specifies the type)
<specific address>%<any zone> (The address specifies the type)
How the unspecified address is treated by the forwarding code is orthogonal
to how applications make use of it to specify which connections or datagrams
are received on a socket. I agree that the forwarding code should treat ::
as a link-local address. I just wanted to point out that the ability to
use the scope type and zone id for filtering could have utility to some class
of applications.
Tim Hartrick
Mentat Inc.
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
