TDD

A test that cannot fail is not a test. The loop is red, green, and then proof: proof that the red came from the missing behavior and the green came from your code.

Everything below is a rule because it has already cost a real campaign.

1. One behavior at a time, vertically

Do not write all the tests and then all the code. Tests written in bulk describe imagined behavior. They assert the shape of things (signatures, data structures) instead of what the system does, so they pass while the feature is broken.

WRONG   RED: test1..test5      GREEN: impl1..impl5
RIGHT   test1 to impl1, then test2 to impl2, then ...