[Nut-upsdev] Re: [nut-commits] svn commit r714 - in trunk: . server

[Arjen de Korte]

> yes, and I suggest to not waste any cycles on trying to have v4-mapped
> addresses matching against v4 policy.

Indeed, we'll leave it in its present state and move the focus to a better
solution. Until then, people will have to choose if they want an IPv4 or
IPv6 listening socket. If one chooses for the latter, IPv4 mapping *may*
work if the OS supports it otherwise, you're out of luck. So far, we've
only supported IPv4, so the addition of the IPv6 code is already somewhat
of an improvement, so I don't think we're seriously hurting anyone here.

The question remains how to migrate to the new situation. If nothing is
specified by the user, we should probably open an AF_INET socket bound to
0.0.0.0 (as has been the default so far in released versions, if nothing
is specified on the commandline). But would it be wise to open an AF_INET6
socket bound to [::] as well, or would that open up a security hole? From
a backwards compatibility point of view, I think we maybe should not open
an AF_INET6 socket anyway and require that this is specified in upsd.conf
by a corresponding LISTEN address, in order not to surprise people not
aware of this change.

Regarding the timeline of IPv6 support, I think we should aim for
introduction with the release of nut-2.2 (provided that's at least a
couple of months from now), not sooner.

Best regards, Arjen
-- 
Eindhoven - The Netherlands
Key fingerprint - 66 4E 03 2C 9D B5 CB 9B  7A FE 7E C1 EE 88 BC 57