[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: IPv6 Link-Local Use Issue for Applications

> Convenience will any day override IETF statements.
> We might be better off if we develop an API through
> which the kernel can inform the applications about
> an ambiguous address. A call to connect() could
> return EAMBIGUOUS. Ofcourse, this error will only
> happen if the user did not specify the scope.

The scope-id field in the sockaddr_in6 is there so that applications can
inform the kernel of the scope of potentially ambiguous addresses.  At
least our implementation (and presumably most others) will not accept a
scoped address in a sockaddr_in6 structure without a properly set
scope-id field.

I really think we're making a mountain out of a molehill here.  Most
apps that use getaddrinfo will naturally support link-local literal
addresses, if the user bothers to type one in.  I suspect mostly only
administrators and the network savvy will ever use literal addresses,
ordinary users will use DNS names.  Link-local addresses won't work in
the DNS, so no one using DNS names will ever get one.  If someone comes
up with a way for ad-hoc applications to present an understandable UI to
ordinary users that happens to use link-local addresses underneath,
well, more power to them.  This is really outside of the IETF's purview
(as actually, is the whole IPv6 API; RFC 2553 and friends are
informational, not standards).

Basically, I agree with Jinmei's points on this issue, so I'll just say
that instead of repeating them.


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 majordomo@sunroof.eng.sun.com