Wherein we write a naive implementation of a test runner, reporter, and assertion API in order to demonstrate how non-magical testing is.

expect-unit.js

/**
 * How to test:
 * 
 * (1) You provide an expression
 * (2) You provide what you "expect" that expression's result/outcome will be
 * (3) You express how you'd like to compare "expected" outcome to the actual outcome
 * (4) Your test should fail initially because you haven't yet writtne the code to support the positive outcome
 * (5) Refactor until the test passes
 * (6) Repeat
 */
var expect, totalTests = 0, passed = 0, failed = 0;

expect = function (actual) {
  var predicate = function (expression){
    totalTests++;
    return (expression) ? ++passed : ++failed;
  }

  // assertion api (a better API would throw exceptions, but let's keep this on topic shall we)
  return {
    toEqual:    function (expected) { return predicate(actual === expected); },
    toNotEqual: function (expected) { return predicate(actual !== expected); }
  };
};

expect(1).toEqual(1);
expect(2).toEqual(2);
expect(8).toEqual(9);

console.log('');
console.log('Total number of tests:', totalTests);
console.log('passed:', passed);
console.log('failed:', failed);