In your previous mail you wrote:
>We could implement Steve's proposal if we invented unique space of
>identifiers that covers all scopes. And I admit it would be useful if
>we can specify the outgoing interface by the sin6_scope_id field
>without any other APIs (such as socket options), but it would be too
>complecated for application programmers, especially on properly using
>different type of sin6_scope_id values (i.e. sometimes use it as an
>interface identifier, and sometimes use it as a zone identifier of a
>larger scope).
Yes, I think I am coming to the same conclusion.
=> what about both:
- put the zone different from the address of the zone stuff into the
new advanced API document (I've already seen this idea in Steve's answer,
we should get the opinion of the new advanced API author(s) and of
course change the name of this thread)
- introduce a flag which allows the extension to different zones
(with the default value, an error should be returned when zones don't
match, a per-user sysctl() should make the job on BSDs)
By the way, I think the interface index confusion is the main problem,
then we should find solutions to this issue, pick one and re-evaluate
the model.
Regards
[EMAIL PROTECTED]
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
