Unit Testing

​The gist of testing

• Programming introduces bugs

• Testing helps find bugs

• Debugging finds and fixes bugs

• Many types of testing: different levels of abstraction and methodologies

• Testing measures code correctness/quality

  • You cannot test correctness/quality into code

• You usually can’t test all possible inputs or even all code paths

  • Aim to get “enough” code coverage. “Enough” is about 85% - 90%.

  • Quality over quantity

• TDD: Write tests first, then write code to pass the tests

  • Tests are formal expressions of the requirements

Unit Testing

• Unit := “smallest testable part of software”

  • Usually few inputs and a single output • Unit Test: Test a single unit (in isolation)

• Goal: validate that each unit is correct

  • If units are correct, they can be (correctly) combined to solve problems

• Typically automated • Many frameworks exist. Mimir uses Google Test for C++

• TDD: write the tests before you write the function

// compute the greatest common divisor
int gcd(int a, int b);
// test
int a = 45, b = 63;
int expected = 9;
int actual = gcd(a, b);
if (actual != expected) {
    cout << “FAIL” << endl;
    cout << “ expected: ” << expected << endl;
    cout << “ got: ” << actual << endl;
} else {
    cout << “PASS” << endl;
}