Panic assertion

The panic assertion checks if a specific piece of code panics:

// Ensure that the code panics
assert_panics!(
    || {
        panic!();
    }
);

// Ensure that the code panics with a specific message
assert_panics!(
    || {
        panic!("hello, world");
    },
    on_message = |message| {
        assert_eq!(message, "hello, world");
    }
);

The second call to assert_panics! takes an on_message argument. This argument is a closure that takes a single argument of type String representing the panic message.

Why not #[should_panic]?

Rust has a built-in attribute called #[should_panic] that can be used to check if a unit test will panic:

#[test]
#[should_panic]
fn unit_test() {
    panic!();

    // Code here will not run
}

There's no reason not to use this, but assert_panics!(...) provides a bit more granularity and can allow code to continue after the panic:

#[test]
fn unit_test() {
    assert_panics!(
        || {
            panic!();
        }
    );

    // This code will still run
}