Don’t Look Back

It’s almost certainly not worth our time and effort to cover an entire complex legacy application before writing any code. We love tests, but the risks involved in doing that much coverage work at once are high, especially if the customer expects us to start working on new functionality.

We draw a line in the sand and start working in a test-driven mode moving forward. One critical element is to ensure that every bug fix starts by writing a failing test somewhere, whether unit, functional, or integration. This is a good way to ramp up tests on our project. It allows us to organically build test coverage over time with relatively small risks to our deadlines and a fairly little chance of breaking existing functionality.