Use a more bulletproof test for whether finite() and isinf() are present.

It seems that recent gcc versions can optimize away calls to these functions
even when the functions do not exist on the platform, resulting in a bogus
positive result.  Avoid this by using a non-constant argument and ensuring
that the function result is not simply discarded.  Per report from
François Laupretre.

1 files changed, 11 insertions, 7 deletions

diff --git a/configure.in b/configure.in
index 5f7acd1871f..9ede74a7b43 100644
--- a/configure.in
+++ b/configure.in
@@ -1,5 +1,5 @@
 dnl Process this file with autoconf to produce a configure script.
-dnl $PostgreSQL: pgsql/configure.in,v 1.441 2006/01/05 03:01:32 momjian Exp $
+dnl $PostgreSQL: pgsql/configure.in,v 1.442 2006/01/12 19:23:22 tgl Exp $
 dnl
 dnl Developers, please strive to achieve this order:
 dnl
@@ -892,12 +892,13 @@ fi
 AC_CHECK_DECLS([snprintf, vsnprintf])
 
 
-# do this one the hard way in case isinf() is a macro
+dnl Cannot use AC_CHECK_FUNC because isinf may be a macro
 AC_CACHE_CHECK([for isinf], ac_cv_func_isinf,
-[AC_TRY_LINK(
-[#include <math.h>
+[AC_TRY_LINK([
+#include <math.h>
+double glob_double;
 ],
-[double x = 0.0; int res = isinf(x);],
+[return isinf(glob_double) ? 0 : 1;],
 [ac_cv_func_isinf=yes],
 [ac_cv_func_isinf=no])])
 
@@ -963,8 +964,11 @@ fi
 
 dnl Cannot use AC_CHECK_FUNC because finite may be a macro
 AC_MSG_CHECKING(for finite)
-AC_TRY_LINK([#include <math.h>],
-            [int dummy=finite(1.0);],
+AC_TRY_LINK([
+#include <math.h>
+double glob_double;
+],
+            [return finite(glob_double) ? 0 : 1;],
             [AC_DEFINE(HAVE_FINITE, 1, [Define to 1 if you have finite().])
 AC_MSG_RESULT(yes)],
             [AC_MSG_RESULT(no)])