@@ -536,31 +536,7 @@ def dec(*args, **kwargs):
 def _have_ieee_doubles():
     if sys.implementation.name == 'cpython':
         return True
-    import math
-    import struct
-    # Check parameters for encoding of floats; a quick exit
-    # if they aren't same as for IEC 60559 doubles.  Check
-    # also that subnormals are present.
-    if (struct.calcsize('d') != 8
-            or sys.float_info.radix != 2
-            or sys.float_info.mant_dig != 53
-            or sys.float_info.dig != 15
-            or sys.float_info.min_exp != -1021
-            or sys.float_info.min_10_exp != -307
-            or sys.float_info.max_exp != 1024
-            or sys.float_info.max_10_exp != 308
-            or not math.issubnormal(math.nextafter(0, 1))):
-        return False
-    # We attempt to determine if this machine is using IEC
-    # floating-point formats by peering at the bits of some
-    # carefully chosen value.  Assume that integer and
-    # floating-point types have same endianness.
-    d = 9006104071832581.0
-    d_be_bytes = b"\x43\x3f\xff\x01\x02\x03\x04\x05"
-    d_packed = struct.pack('d', d)
-    if sys.byteorder == 'little':
-        return d_packed == bytes(reversed(d_be_bytes))
-    return d_packed == d_be_bytes
+    return False
 
 HAVE_IEEE_754 = _have_ieee_doubles()