Peter Eisentraut wrote:
> Bruce Momjian writes:
>
> > Peter was the first to mention it. His reasoning was that if IPv6 was
> > working, but then stopped working, the admin would never know on startup
> > because of the IPv4 fallback.
>
> My view was that we should treat unix, ipv4, and ipv6 as independent
> address families each with their own on/off switch (except that unix
> doesn't have an off switch). Tom's view is that we should treat ipv4 and
> ipv6 as effectively one address family. That makes sense, too, and it is
> probably more with the spirit of IPv6.
OK, Peter, to keep you and everyone happy, what changes are your
proposing to the existing code, if any. The only current behavior is
printing an IPv6 failure for IPv6-enabled backend in the server logs.
-- Bruce Momjian | http://candle.pha.pa.us pgman@candle.pha.pa.us | (610)
359-1001+ If your life is a hard drive, | 13 Roberts Road + Christ can be your backup. | Newtown Square,
Pennsylvania19073