Generated files problems.

[Steve Kargl]

I'm trying to remove the inlining of rrspacing by 
calling library routines generated from a rrspacing.m4
file.  I wasted several hours on this task, and 
mounting frustration has taken its toll.

In a moment of pure genius, I revert my entire tree
to stock trunk.  There are no local changes in my 
tree.  I then did

cd trunk/libgfortran
rm generated/fraction_r*
cd ../../obj
../trunk/configure --disable-libmudflap --disable-intl \
   --enable-languages=c,fortran --enable-maintainer-mode
   --prefix=$HOME/work/4x
gmake bootstrap

This completes the build.  However, upon inspection of the
newly generated fraction_r4.c file via
"svn diff generated/fraction_r4.c", I find


Index: generated/fraction_r4.c
===================================================================
--- generated/fraction_r4.c	(revision 117447)
+++ generated/fraction_r4.c	(working copy)
@@ -33,16 +33,16 @@ Boston, MA 02110-1301, USA.  */
 #include "libgfortran.h"
 
 
-#if defined (HAVE_GFC_REAL_4) && defined (HAVE_FREXPF)
+#if defined (HAVE_GFC_REAL_) && defined (HAVE_FREXP_)
 
-extern GFC_REAL_4 fraction_r4 (GFC_REAL_4 s);
-export_proto(fraction_r4);
+extern GFC_REAL_ fraction_r (GFC_REAL_ s);
+export_proto(fraction_r);
 
-GFC_REAL_4
-fraction_r4 (GFC_REAL_4 s)
+GFC_REAL_
+fraction_r (GFC_REAL_ s)
 {
   int dummy_exp;
-  return frexpf (s, &dummy_exp);
+  return frexp_ (s, &dummy_exp);
 }
 
 #endif

which clearly is fucked up.  So, if we have any m4/file
generation experts out there.  I'd appreciate a pointer.

-- 
Steve