Ludovic Courtès wrote: > I think the problem is that ‘gl_HEADER_NETDB’ determines ‘NETDB_H=''’ > because <netdb.h> is indeed available and it does define ‘struct > addrinfo’; it just happens to lack some of the AI_ macros.
That explains it. I propose to generate netdb.h always, like we do with stdlib.h and unistd.h for 3 years and for many others for 2 months. Then we can also add GNULIB_POSIXCHECK advice to <netdb.h>. Simon, is that ok with you? 2010-04-16 Bruno Haible <[email protected]> Override netdb.h always. * modules/netdb (Makefile.am): Augment BUILT_SOURCES always. * m4/netdb_h.m4 (gl_HEADER_NETDB): Don't set NETDB_H. Reported by Ludovic Courtès <[email protected]>. --- m4/netdb_h.m4.orig Fri Apr 16 12:05:25 2010 +++ m4/netdb_h.m4 Fri Apr 16 12:03:21 2010 @@ -1,4 +1,4 @@ -# netdb_h.m4 serial 7 +# netdb_h.m4 serial 8 dnl Copyright (C) 2008, 2009, 2010 Free Software Foundation, Inc. dnl This file is free software; the Free Software Foundation dnl gives unlimited permission to copy and/or distribute it, @@ -10,21 +10,11 @@ AC_CHECK_HEADERS_ONCE([netdb.h]) gl_CHECK_NEXT_HEADERS([netdb.h]) if test $ac_cv_header_netdb_h = yes; then - AC_COMPILE_IFELSE( - [AC_LANG_PROGRAM([[ - #include <netdb.h> - struct addrinfo a; - int b = EAI_OVERFLOW; - int c = AI_NUMERICSERV; - ]])], - [NETDB_H=''], [NETDB_H='netdb.h']) HAVE_NETDB_H=1 else - NETDB_H='netdb.h' HAVE_NETDB_H=0 fi AC_SUBST([HAVE_NETDB_H]) - AC_SUBST([NETDB_H]) ]) AC_DEFUN([gl_NETDB_MODULE_INDICATOR], --- modules/netdb.orig Fri Apr 16 12:05:25 2010 +++ modules/netdb Fri Apr 16 12:03:28 2010 @@ -14,7 +14,7 @@ gl_HEADER_NETDB Makefile.am: -BUILT_SOURCES += $(NETDB_H) +BUILT_SOURCES += netdb.h # We need the following in order to create <netdb.h> when the system # doesn't have one that works with the given compiler.
