Add Float32Array double-to-float cast testcase

tests/ecmascript/test-bi-typedarray-double-to-float.js

@@ -0,0 +1,43 @@
+/*
+ *  Exercise double-to-float out-of-range behavior which is technically
+ *  undefined behavior in C99.
+ */
+
+/*===
+3.4028234663852886e+38
+3.4028234663852886e+38
+Infinity
+-3.4028234663852886e+38
+-3.4028234663852886e+38
+-Infinity
+done
+===*/
+
+var f32 = new Float32Array(1);
+
+/* Maximum float (as double). */
+f32[0] = 340282346638528859811704183484516925440.0;
+print(f32[0]);
+
+/* Maximum double that rounds to maximum float (rather than infinity).
+ * Without explicit handling for the out-of-range (out of 'float' range
+ * here) case, -fsanitize=undefined will prompt:
+ * duk_util_cast.c:139:11: runtime error: value 3.40282e+38 is outside the range of representable values of type 'float'
+ * Same for all casts where the argument is beyond 'float' range.
+ */
+f32[0] = 340282356779733623858607532500980858880.0;
+print(f32[0]);
+
+/* First double that rounds to infinity. */
+f32[0] = 340282356779733661637539395458142568448.0;
+print(f32[0]);
+
+/* And same as negative. */
+f32[0] = -340282346638528859811704183484516925440.0;
+print(f32[0]);
+f32[0] = -340282356779733623858607532500980858880.0;
+print(f32[0]);
+f32[0] = -340282356779733661637539395458142568448.0;
+print(f32[0]);
+
+print('done');