implement bool conversion for float

jit/src/instructions.rs

@@ -82,7 +82,11 @@ impl<'a, 'b> FunctionCompiler<'a, 'b> {
 
     fn boolean_val(&mut self, val: JitValue) -> Result<Value, JitCompileError> {
         match val.ty {
-            JitType::Float => Err(JitCompileError::NotSupported),
+            JitType::Float => {
+                let zero = self.builder.ins().f64const(0);
+                let val = self.builder.ins().fcmp(FloatCC::NotEqual, val.val, zero);
+                Ok(self.builder.ins().bint(types::I64, val))
+            },
             JitType::Int => Ok(val.val),
         }
     }

jit/tests/float_tests.rs

@@ -90,3 +90,20 @@ fn test_div() {
     assert_bits_eq!(div(2.0, f64::NEG_INFINITY), Ok(-0.0f64));
     assert_bits_eq!(div(-1.0, f64::INFINITY), Ok(-0.0f64));
 }
+
+#[test]
+fn test_if_bool() {
+    let if_bool = jit_function! { if_bool(a:f64) -> i64 => r##"
+        def if_bool(a: float):
+            if a:
+                return 1
+            return 0
+    "## };
+
+    assert_eq!(if_bool(5.2), Ok(1));
+    assert_eq!(if_bool(-3.4), Ok(1));
+    assert_eq!(if_bool(f64::NAN), Ok(1));
+    assert_eq!(if_bool(f64::INFINITY), Ok(1));
+
+    assert_eq!(if_bool(0.0), Ok(0));
+}