> From: JINMEI Tatuya / 神明達哉 <[EMAIL PROTECTED]>
> As described in the 03 draft, one possibility is a kind of MIB entry
> which only specifies a particular zone (of a particular scope type).
> Such a possibility is one of the reasons to adopt this type of format
> since the 03 version.
Ahh, a nasty person might then ask: why are we burdening all other
addresses with this <scope_type>
<address>%<scope_type>.<id_in_the_scope>
when they don't need it, and the only place where it would be needed
(that MIB example), is not actually using address at all!
But, of course I can guess why: the place where such notation occurs,
is usually parsed as address, and it would be truly neat if it could
handle the addressless zone id too.
My proposal for a solution is as follows:
1) <address>%<zoneid> is the format. scope_type is always known from
the <address>.
2) to allow an API to specify an ANY address limited to a specific
scope, the following specic adresses are assigned for the
purpose:
fe80::%<zoneid> - ANY address in specified link local zone
fec0::%<zoneid> - ANY address in specified site zone
::%<zoneid> - ANY address in specified global zone
and just to be complete, for zone specific multicasts
ff01::%<zoneid>
ff02::%<zoneid>
ff03::%<zoneid>
...
ff0F::%<zoneid>
I believe all (?) of above are sort of illegal in normal use, and
could thus be dedicated for the purpose (MIB, limiting a listen socket
to specific scopeid etc.).
How does this kludge sound?
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
