[Bug-gnulib] gl_AC_TYPE_UINT32_T broken

[Albert Chin]

gl_AC_TYPE_UINT32_T doesn't correctly define uint32_t on Solaris
2.5.1/SPARC. It should be obvious why after looking at the patch
below.

-- 
albert chin (address@hidden)

-- snip snip
        * m4/uint32_t.m4: When determining uint32_t equivalent,
        substitute $ac_type for equivalent type rather than
        blindly using uint32_t *always* which won't work if
        uint32_t is unavailable.

Index: m4/uint32_t.m4
===================================================================
RCS file: /cvsroot/coreutils/coreutils/m4/uint32_t.m4,v
retrieving revision 1.2
diff -u -3 -p -r1.2 uint32_t.m4
--- m4/uint32_t.m4      4 Aug 2004 22:30:38 -0000       1.2
+++ m4/uint32_t.m4      3 Nov 2004 04:33:47 -0000
@@ -27,11 +27,11 @@ AC_DEFUN([gl_AC_TYPE_UINT32_T],
        AC_COMPILE_IFELSE(
         [AC_LANG_BOOL_COMPILE_TRY(
            [AC_INCLUDES_DEFAULT],
-           [[(uint32_t) -1 == 4294967295U]])],
+           [[($ac_type) -1 == 4294967295U]])],
         [gl_cv_c_uint32_t=$ac_type])
-       test $gl_cv_c_uint32_t != no && break
+       test "$gl_cv_c_uint32_t" != no && break
      done])
-  case $gl_cv_c_uint32_t in
+  case "$gl_cv_c_uint32_t" in
   no|uint32_t) ;;
   *)
     AC_DEFINE_UNQUOTED(uint32_t, $gl_cv_c_uint32_t,