On 16/04/2014 17:14, Alvaro Herrera wrote:
> Marco Atzeri wrote:
>> On 13/04/2014 18:09, Tom Lane wrote:
>>> Andres Freund <andres@2ndquadrant.com> writes:
>>>> On 2014-04-12 16:35:48 -0400, Tom Lane wrote:
>>>>> In principle, that commit shouldn't have affected behavior for pg_hba
>>>>> entries with numeric address fields ...
>>>
>>>> Hm. getaddrinfo.c has this bit:
>>>> /* Unsupported flags. */
>>>> if (flags & NI_NAMEREQD)
>>>> return EAI_AGAIN;
>>>
>>> Yeah, and that flag is only ever specified when attempting to do reverse
>>> lookup on a client address to see if it matches a non-numeric pg_hba
>>> entry.
>
> I don't know if this is relevant, but perhaps we're defining the
> constants in a way that conflicts with the values defined by cygwin. A
> very quick search finds a 2007 patch for Mutt[1] that seems to have
> NI_NAMEREQD defined as 8 somewhere, while 4 is NI_NOFQDN. But we have
> this in getaddrinfo.h:
>
> #ifndef NI_NAMEREQD
> #define NI_NAMEREQD 4
> #endif
>
> So maybe we're doing something wrong. Indeed, my system has in
> /usr/include/netdb.h
>
> # define NI_NAMEREQD 8 /* Don't return numeric addresses. */
>
> You'd do well to research this more, I think.
>
> [1] http://marc.info/?l=mutt-dev&m=117752314512877&w=2
on cygwin both 32 and 64 bit I see
netdb.h:#define NI_NAMEREQD 0x4 /* Not being able to resolve is
an error. */
same on
w32api/ws2tcpip.h:#define NI_NAMEREQD 0x04
curiosly I see also on
roken-common.h:#define NI_NAMEREQD 0x02
$ cygcheck -f /usr/include/roken-common.h
libkrb5-devel-1.5.3-1
not sure if it has any relevance at all in this case.
>