* acinclude.m4, configure.in (GMP_C_RIGHT_SHIFT): Remove, no longer

	needed.

1 files changed, 0 insertions, 36 deletions

diff --git a/acinclude.m4 b/acinclude.m4
index 73a791138..4ca9d79fc 100644
--- a/acinclude.m4
+++ b/acinclude.m4
@@ -3006,42 +3006,6 @@ esac
 ])
 
 
-dnl  GMP_C_RIGHT_SHIFT
-dnl  -----------------
-dnl  Test whether a signed right shift is arithmetic, ie. high bit positions
-dnl  are filled with copies of the sign bit.
-dnl
-dnl  C99 says a signed right shift of a negative is "implementation
-dnl  defined".  We take that to mean it might be arithmetic or it might not,
-dnl  but that it's consistently one way or another way and not merely
-dnl  random.
-dnl
-dnl  Right shifts are usually arithmetic in C.  Cray vector systems are the
-dnl  only place we know of where this isn't so.  Those systems don't have an
-dnl  arithmetic right shift instruction so the compiler gives a logical
-dnl  shift (ie. high bits filled with zeros) instead.  Or at least this
-dnl  happens for operations on "long"s, but not sure about "int", they might
-dnl  be arithmetic.  Cray alpha systems have a cc -hnosignedshifts option to
-dnl  do the same there too.
-dnl
-dnl  The test here generates a negative array size and hence a compiler
-dnl  error if a right shift doesn't replicate the sign bit.
-
-AC_DEFUN([GMP_C_RIGHT_SHIFT],
-[AC_CACHE_CHECK([whether signed right shifts are arithmetic],
-                gmp_cv_c_right_shift,
-[AC_TRY_COMPILE(
-[int foo [(-1L >> 1) < 0L ? 1 : -1];],,
-gmp_cv_c_right_shift=yes, gmp_cv_c_right_shift=no)
-])
-if test $gmp_cv_c_right_shift = yes; then
-  AC_DEFINE(HAVE_RIGHT_SHIFT_ARITHMETIC, 1,
-            [Define to 1 if signed right shifts (of longs) are "arithmetic",
-             ie. shift in copies of the sign bit.])
-fi
-])
-
-
 dnl  GMP_C_STDARG
 dnl  ------------
 dnl  Test whether to use <stdarg.h> or <varargs.h>.