>IMHO, we should enable ipv4-mapped address support by default
>for RFC compliance. And bind(2) with both of ipv4 and ipv6
>should be accepted if IPV6_V6ONLY has been set to ipv6 socket
>for compatibility with application already, wildly being used
>on Linux Systems.
if I were you I'd propose to drop RFC conformance and disable IPv4
mapped address by default, and have IPV6_V6ONLY socket option for
re-enabling it (= kame/netbsd behavior). when default bevhavior is
on insecure side I get worried.
another issue is bind(2) ordering and conflict table... :-P
>I'd like to talk about this issue with itojun, jinmei and others.
>In IETF?
of course. i heard that you will have a linux-usagi unofficial
meeting at th venue, i'll try to show up there.
itojun
