Boundaries & Trustworthy Tests

Two habits separate throwaway tests from trustworthy ones.

Test the edges, not just the happy path

def test_grade():
    assert letter_grade(90) == "A"    # the boundary
    assert letter_grade(89.9) == "B"  # just below it — bugs live here
    assert letter_grade(150) == "A"   # nonsense input, defined behavior

Bugs cluster at boundaries: zero, empty, one-past-the-end, the day the discount starts. Happy-path-only tests are passports that wave everything through.

A failing test must fail for the RIGHT reason

Before celebrating a test that fails, read the error. Is it failing because the code is wrong, or because the test itself is wrong? A test suite nobody trusts is worse than no suite — it prints false alarms until people stop listening.