Merge pull request #2085 from bbbradsmith/numerical_constant_errors-float

Numerical constant errors and improvements (float)

test/val/float-const-convert.c

@@ -0,0 +1,14 @@
+/* Demonstrates that floating point constants are allowed in a limited way.
+   Value will be converted to an int, with a warning if precision is lost. */
+
+int a = 3.0;
+int b = 23.1;
+int c = -5.0;
+
+int main(void)
+{
+    if (a != 3) return 1;
+    if (b != 23) return 2;
+    if (c != -5) return 3;
+    return 0;
+}