As illustrated in the preceding section, code coverage tools can help you uncover mistakes in your code. However, they should be used as a diagnostic tool only; you shouldn't be chasing after 100% code coverage as a goal in itself.

This is because code coverage has no relation to the quality of your tests. You can define test cases that cover 100% of your code, but if the assertions are wrong, or if the tests have errors in it, then the perfect coverage means nothing. For instance, the following test block will always pass, even though one of the assertions suggests it would fail:

it('This will always pass', function () {
  it('Even though you may expect it to fail', function () {
    assert(true, false);
  });
});

This highlights the point that code coverage cannot detect bad tests. Instead, you should focus on writing meaningful tests that will actually show bugs when they arise; if you do that, the test coverage will naturally remain high, and you can use the reports to improve whatever you've missed in your tests.