Peter Eisentraut wrote:
> Bruce Momjian writes:
>
> > The IPv6 patch currently checks for the function getaddrinfo() and the
> > include file netinet/ip6.h.
>
> getaddrinfo() has nothing to do with IPv6, and netinet/ip6.h isn't
> included anywhere, so why check for it?
>
> I believe we would need to check for a combination of
>
> getaddrinfo()
> inet_ntop()/inet_pton()
> AF_INET6
> struct sockaddr_in6
>
> Then again, it might be simpler to make it a configure option
> (--enable-ipv6) and just assume all those things exist. Over time the
> interfaces might stabilize.
>
> However, even when IPv6 support is compiled in, we still need some
> run-time control about whether it should get activated because the
> presence of the programming interfaces doesn't tell you anything about
> whether the run-time environment supports it.
>
I changed the check for sockaddr_in6, rather than ip6.h. Let's see how
that works. I would like to have this automatic rather than adding yet
another configure flag.
-- 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