Fix issue #742 (#1650)

It is a workaround fix. The problem comes from the inaccuracy of the double rounding.

JerryScript-DCO-1.0-Signed-off-by: Robert Sipka rsipka.uszeged@partner.samsung.com

jerry-core/ecma/builtin-objects/ecma-builtin-number-prototype.c

@@ -253,10 +253,15 @@ ecma_builtin_number_prototype_object_to_string (ecma_value_t this_arg, /**< this
         scale = -scale;
       }
 
-      int buff_size;
+      int buff_size = 1;
       if (is_scale_negative)
       {
-        buff_size = (int) floor (log (this_arg_number) / log (radix)) + 1;
+        double counter = this_arg_number;
+        while (counter > radix)
+        {
+          counter /= radix;
+          buff_size++;
+        }
       }
       else
       {

tests/jerry/number-prototype-to-string.js

@@ -80,6 +80,12 @@ assert((123400).toString(34) === "34pe");
 assert((123400).toString(35) === "2upp");
 assert((123400).toString(36) === "2n7s");
 
+assert ((1152921504606846600).toString([1,2,3,4].slice(1,2)) === "111111111111111111111111111111111111111111111111111010000000");
+assert ((-1152921504606846600).toString(2) === "-111111111111111111111111111111111111111111111111111010000000");
+
+assert ((0x100000000000061).toString(2) === "100000000000000000000000000000000000000000000000001100000")
+assert ((-0x100000000000061).toString(16) === "-100000000000060");
+
 assert((123400).toString(new Number(16)) === "1e208");
 
 var digit_chars = ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9',