> I have some comments about the scoped @ arch, about the scope_id.
> - I definitly agree with E. Nordmark about storing all space names in a
> single 32 bit box! This could be done using a quartet with same
> signification as for multicast. Scope id could look like something as
> 0x 00 0S zone_id
> with 16 bits for zone id, and S being set to 2 for link-id, 5 for
> site-id ...
> zone_id might be set to nul to let the default zone ot be taken. hence
> 0x00020000 being the default link.
For compatibility with what we have in RFC 2553 I think you want to
still handle just an ifindex i.e. the high order bits being zero should
mean that the low order bits is an ifindex.
But I still think your general approach with using the defined (multicast)
scope values is a good idea. Of course, this can all be hidden in
the if_nametoindex type APIs.
Thus you'd get
0 Interface ID
1 -
2 Link ID
5 Site ID
...
If we go this path (allowing different scope zones ids in sin6_scope_id)
it might make sense to add parallels to the if_nametoindex() etc APIs.
Either
site_nametoindex(), site_indextoname(), site_nameindex()
link_nametoindex(), ...
or a more general set of routines
sz_nametoindex(char *name, int scope_level)
That way we can add protocol support (something like MZAP?)
for advertising scope zone names in the future.
Erik
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
