Bruno Haible <[EMAIL PROTECTED]> writes:
> Eric Blake wrote:
>> Won't work on cygwin.
>> ...
>> Checking for ws2tcpip.h is problematic. Instead, you need to reuse the
>> check
>> made in m4/sys_socket_h.m4, since ws2tcpip.h is present but cannot be
>> compiled
>> on cygwin.
>
> Ah, right. I h
Eric Blake wrote:
> Won't work on cygwin.
> ...
> Checking for ws2tcpip.h is problematic. Instead, you need to reuse the check
> made in m4/sys_socket_h.m4, since ws2tcpip.h is present but cannot be
> compiled
> on cygwin.
Ah, right. I had forgotten about it. Thanks for catching this!!
How ab
Bruno Haible clisp.org> writes:
> How about this?
>
> 2007-06-27 Simon Josefsson josefsson.org>
> Bruno Haible clisp.org>
>
> * m4/getaddrinfo.m4 (gl_GETADDRINFO): Check for ws2tcpip.h before
> using HAVE_WS2TCPIP_H. Check for gai_strerror through an explicit
>
Hi Simon,
> - AC_REPLACE_FUNCS(gai_strerror)
> + # We can't use AC_REPLACE_FUNCS here because gai_strerror may be an
> + # inline function declared in ws2tcpip.h, so we need to get that
> + # header included somehow.
> + AC_MSG_CHECKING([for gai_strerror (possibly via ws2tcpip.h)])
> + AC_TR
Hi! Mingw32 contains in ws2tcpip.h:
static __inline char*
gai_strerrorA(int ecode)
...
#define gai_strerror gai_strerrorA
Thus, the following test in getaddrinfo.m4 will incorrectly return 'no':
AC_REPLACE_FUNCS(gai_strerror)
since the function isn't in any library.
How about the following
