Tom Lane wrote:
>Andrew Dunstan <andrew@dunslane.net> writes:
>
>
>>Try this instead if you prefer.
>>
>>
>
>I thought this was a little bit too trusting about there being a
>getaddrinfo to probe, so I tightened it up as per the attached
>applied patch.
>
>Where are we at this point on the Windows/IPv6 issue --- are there
>more fixes to come, or is it done?
>
>
Not done yet. One thing left.
The idea was that we would put dynamic testing of properly working
getaddrinfo and friends on Windows into our getaddrinfo.c. That would
be the "local tweak" you mentioned previously ;-)
In consequence of that plan, I think we would need to remove "&&
defined(HAVE_GETADDRINFO)" from your applied patch and let it fall
through to our homegrown getaddrinfo if there isn't one. On most such
platforms it would fail, consuming a few more millisecs, but with
Windows with the expected patch it could pass.
(I know it's a muddle - I can't think how we came not to do IPv6 for
Windows in 8.0, or at the very least put it on the TODO list.)
cheers
andrew