Improve `as` casting exercise.

diff --git a/exercises/02_basic_calculator/10_as_casting/src/lib.rs b/exercises/02_basic_calculator/10_as_casting/src/lib.rs
index 7b8b3e0..b1921d0 100644 (file)
--- a/exercises/02_basic_calculator/10_as_casting/src/lib.rs
+++ b/exercises/02_basic_calculator/10_as_casting/src/lib.rs
@@ -6,17 +6,32 @@ mod tests {
 
     #[test]
     fn u16_to_u32() {
-        assert_eq!(47u16 as u32, todo!());
+        let v: u32 = todo!();
+        assert_eq!(47u16 as u32, v);
     }
 
     #[test]
-    #[allow(overflowing_literals)]
     fn u8_to_i8() {
-        assert_eq!(255 as i8, todo!());
+        // The compiler is smart enough to know that the value 255 cannot fit
+        // inside an i8, so it'll emit a hard error. We intentionally disable
+        // this guardrail to make this (bad) conversion possible.
+        // The compiler is only able to pick on this because the value is a
+        // literal. If we were to use a variable, the compiler wouldn't be able to
+        // catch this at compile time.
+        #[allow(overflowing_literals)]
+        let x = { 255 as i8 };
+
+        // You could solve this by using exactly the same expression as above,
+        // but that would defeat the purpose of the exercise. Instead, use a genuine
+        // `i8` value that is equivalent to `255` when converted from `u8`.
+        let y: i8 = todo!();
+
+        assert_eq!(x, y);
     }
 
     #[test]
     fn bool_to_u8() {
-        assert_eq!(true as u8, todo!());
+        let v: u8 = todo!();
+        assert_eq!(true as u8, v);
     }
 }