Testing

We can prefix a function with the #[test] attribute to indicate that it is part of the unit tests for our application or library. We then compile with the following command and run the generated executable:

rustc --test program.rs

This will replace the main() function with a test runner, and show the result from the functions marked with #[test], for example:

// from Chapter 3/code/attributes_testing.rs 
fn main() { 
  println!("No tests are compiled,compile with rustc --test! "); 
} 
 
#[test] 
fn arithmetic() { 
  if 2 + 3 == 5 { 
    println!("You can calculate!"); 
  } 
} 

Test functions, like arithmetic() in the example, are black boxes, they have no arguments or returns. When this program is run on the command line it produces the following output:

But, if we change the test to if 2 + 3 == 6 the test passes as well! Try it out. It turns out that test functions always pass when their execution does not cause a crash (called a panic in Rust terminology), and only fails when it panics...