F7.1 · LESSON 1 OF 6

Tests as examples

Build a small HTTP service with clear boundaries, tests, logs and a repeatable runtime.

THE IDEA

A test is an example with a verdict

A test is a tiny program that calls your code with a specific input and states what the result must be. It is an example that can fail. Written before or after the code, a good set of tests is also a specification: anyone can read them and learn what the function promises, including the awkward corners.

A plain assert, and what a failure looks like (pytest style)

def test_average_of_two():
    assert average([1, 2]) == 1.5

# if average used // instead of /, pytest reports roughly:
>       assert average([1, 2]) == 1.5
E       assert 1 == 1.5
E        +  where 1 = average([1, 2])

A failing test message should tell you three things: which example, what you expected, and what you got. That is why tests with a clear name and one idea each are easier to act on than one giant test.

Easy examples miss bugs

Imagine average(values) must return the mean as a float, and None for an empty list (there is no mean of nothing, and dividing by a count of zero is an error). Below are seven broken copies, each with one realistic bug. A test only has value if some example makes a broken copy fail. Switch examples on and watch which copies get caught.

Example tests (click to switch on)

Broken copy of average	c1	c2	c3	c4	c5	c6	c7	Verdict
divides by count - 1	FAIL	would fail	would fail	would fail	would fail	pass	would fail	caught
returns 0 for an empty list instead of None	pass	pass	pass	pass	pass	would fail	pass	survives: nobody noticed
rounds the mean down (uses // instead of /)	pass	pass	would fail	would fail	pass	pass	pass	survives: nobody noticed
ignores the last value	FAIL	would fail	would fail	would fail	would fail	pass	would fail	caught
takes the absolute value of every number	pass	pass	pass	pass	would fail	pass	pass	survives: nobody noticed
rounds the mean to one decimal place	pass	pass	pass	would fail	pass	pass	pass	survives: nobody noticed
mishandles a list with exactly one value	pass	would fail	pass	pass	pass	pass	pass	survives: nobody noticed

5 broken copies still pass your tests. Notice that c7 adds nothing once c1 is on: a test earns its place by catching something the others do not.

Start with c1 alone: it catches only some bugs. Add the boundary cases (empty list, single value) and the awkward ones (a mean that is not whole, a negative number).

Which cases matter

Typical: a normal input with an answer you can verify by hand.
Boundaries: empty, one element, zero, the largest allowed value.
Different kinds of input: negatives, repeated values, results that are not whole numbers.
Failure: what the function does when the input is not acceptable. State it, then test it.

Predict first. A function divides a total by a count. Which extra case is most important to test?

Chapter context

Engineering makes a program understandable when it fails or changes. Separate pure calculations from external I/O so behavior can be checked cheaply and errors can be connected to specific inputs.

TRY IT YOURSELF

Exercise: write tests that catch seven broken copies

average(values) is correct below. A hidden checker also keeps **seven broken copies** of it (each has one realistic bug) and runs your test_average against every one.
Write assert lines inside test_average(average). Use the **parameter** average, not the one defined above it, so the checker can swap in the broken copies.
A broken copy is caught when one of your asserts fails (or crashes) on it. The correct function must pass all of your asserts, so every expected value you write has to be right.
Each result line names the assert that stopped the broken copy, so you can see which example earns its place.

def average(values):
    """Mean of a list of numbers, as a float. An empty list has no mean: return None."""
    if not values:
        return None
    return sum(values) / len(values)


def test_average(average):
    # 'average' is passed in so the checker can run this test against other versions of it.
    # Write assert statements here, for example:
    #   assert average([2, 4]) == 3.0
    pass

Ctrl/⌘ + Enter runs it

Stuck, or something not working? You can come back to it any time.

CHECK YOUR UNDERSTANDING

What extra case matters for a function dividing by a count?

AA total of exactly 100, because round numbers hide rounding bugs.BA zero count, because the normal arithmetic is undefined.CA very large total, because big numbers overflow in Python.

Chapter reference: math, worked example & code

This section belongs to the whole chapter, so it is the same on every lesson in it and may cover a different lesson's topic. The code for the experiment you just ran is inside the experiment itself.

Math and notation

For money examples, use integer minor units rather than binary floating point. The subtotal is sum(quantity × unit_price), and the domain must specify rounding and tax rules before those operations are implemented.

Worked example

Two items cost 125 minor units each and three cost 40 each. The subtotal is 2×125 + 3×40 = 370. A test can exercise this calculation without starting an HTTP server. Request parsing and response formatting need separate checks.

Reference code

The following example uses the Python standard library and does not call a remote service.

def subtotal(items):
    return sum(quantity * price for quantity, price in items)
assert subtotal([(2, 125), (3, 40)]) == 370
assert subtotal([]) == 0

Python examples are provided for study and local execution; this page does not execute Python.

Common mistakes

Test coverage counts executed lines, not verified requirements. A container does not turn an unsafe operation into a safe one. JSON validity alone does not establish application validity.