On 01/03/2011 12:15 PM, I wrote:
The following patch allows me to build the 8.3 and 8.4 branches using
Visual Studio 2008, once the build system is patched. But I don't
really know why. HEAD and 9.0 build fine without it. But those
branches branches fail with a complaint about IPPROTO_IPV6 being
undefined.
The patch seems harmless enough. But I'd like to know why it's
happening. Does anyone have a clue?
-#ifdef IPV6_V6ONLY
+#if defined(IPV6_V6ONLY) && defined(IPPROTO_IPV6)
if (addr->ai_family == AF_INET6)
{
if (setsockopt(fd, IPPROTO_IPV6, IPV6_V6ONLY,
OK, what's going here is that, in the newer SDK, IPV6_V6ONLY is defined
unconditionally, but IPPROTO_IPV6 is only defined if _WIN32_WINNT is set
to 0x0501 or higher. We defined _WIN32_WINNT as 0x0500 until 9.0, when
we changed it specifically to allow use of the right IPV6 settings.
This seems to me like a clear error in the MS headers. I don't think it
makes any sense to define the settings constant but not the context
constant. The fix I have suggested above doesn't seem unreasonable or
terribly unsafe in these circumstances. The code clearly contemplates
the setsockopt() call in question not having been run, as shown in this
comment:
/*
* Note: This might fail on some OS's, like Linux older than
* 2.4.21-pre3, that don't have the IPV6_V6ONLY socket option,
and map
* ipv4 addresses to ipv6. It will show ::ffff:ipv4 for all ipv4
* connections.
*/
cheers
andrew
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers