Testing

Testing

Good tests pin down behavior and fail for one clear reason. Chase meaningful cases, not a coverage number.

What to test (priority order)

1. The contract — for each function: typical input → expected output.
2. Edge cases — empty, single element, very large, zero/negative, null/None, unicode, boundary values (off-by-one), duplicates.
3. Error paths — invalid input rejected, exceptions raised/handled, partial failure.
4. Regressions — every fixed bug gets a test that fails before the fix and passes after.
5. Integration seams — where modules/services meet (fewer, broader tests here).

Shape: Arrange–Act–Assert

def test_discount_caps_at_zero():
    cart = Cart(items=[Item(price=10)])   # arrange
    total = cart.total(discount=2.0)      # act  (200% off)
    assert total == 0                     # assert: never negative

• One behavior per test; name it after the behavior (test_discount_caps_at_zero).
• Make failures readable: assert on specific values, not just "truthy".

Practical guidance

• Prefer many small fast unit tests + a few integration tests (the test pyramid).
• Keep tests deterministic: no real network/clock/randomness — inject or mock them.
• Mock at boundaries you own; don't mock the thing under test.
• For a bug: write the minimal failing test first, then fix until it's green.
• Run the suite to confirm, and report what passed/failed — don't claim green without running it.