Re: Building universal binaries makes 'check' fail

[Peter O'Gorman]

On Mar 6, 2007, at 9:50 AM, Paul Eggert wrote:

> "Peter O'Gorman" <address@hidden> writes:
>
> > What if the package does not use AC_CONFIG_HEADERS? This patch will
> > fail. What about AC_CHECK_SIZEOF which will report incorrect results
> > if -arch i386 -arch x86_64 are specified for example?
>
> Those problems existed in the previous Autoconf version too, so the
> patch shouldn't made things worse.  Since the patch does fix the
> problem for coreutils on a real platform it seems like it's a win.
>
> We can solve the other problems later, as needed.
 [cut bug-gnulib and bug-coreutils]

Hi Paul,
Okay, but can we maybe solve this one in a way that does not cause  
regressions for the much more usual case of building thin. Right now,  
if there is not a config header in use, on darwin, while we used to  
get -DWORDS_BIGENDIAN=1 on the compile line for ppc, with current cvs  
we get nothing at all. There are quite a number of software packages  
out there that do not use config headers for whatever reason.

Here is a suggested patch.

Peter


--- lib/autoconf/c.m4   1 Mar 2007 00:18:28 -0000       1.244
+++ lib/autoconf/c.m4   6 Mar 2007 04:01:26 -0000
@@ -1381,6 +1381,7 @@ AC_DEFUN([AC_C_BIGENDIAN],
#endif])dnl
  AC_CACHE_CHECK([whether byte ordering is bigendian],  
[ac_cv_c_bigendian],
    [ac_cv_c_bigendian=unknown
+    m4_ifdef([AH_HEADER],[
     # See if __BIG_ENDIAN__ or __LITTLE_ENDIAN__ is defined.
     AC_COMPILE_IFELSE(
       [AC_LANG_SOURCE(
@@ -1389,7 +1390,7 @@ AC_DEFUN([AC_C_BIGENDIAN],
          #endif
          typedef int dummy;
         ]])],
-      [ac_cv_c_bigendian=universal])
+      [ac_cv_c_bigendian=universal])])
     if test $ac_cv_c_bigendian = unknown; then
       # See if sys/param.h defines the BYTE_ORDER macro.
       AC_COMPILE_IFELSE(