Test to demonstrate availability of floating point constants, document the possibility.

doc/cc65.sgml

@@ -805,6 +805,9 @@ and the one defined by the ISO standard:
 <itemize>
 
 <item>  The datatypes "float" and "double" are not available.
+        Floating point constants may be used, though they will have to be
+        converted and stored into integer values.
+        Floating point arithmetic expressions are not supported.
         <p>
 <item>  C Functions may not return structs (or unions), and structs may not
         be passed as parameters by value. However, struct assignment *is*

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;
+}