To TDD or not to TDD?

Skeptical readers may be thinking, “I can get the same benefits without doing TDD. I can think more about my requirements, force myself to only implement what is needed, and consider the testability of my class from the beginning. I do not need to write tests for that!” That is true. But I appreciate TDD because it gives me a rhythm to follow. Finding the next-simplest feature, writing a test for it, implementing nothing more than what is needed, and reflecting on what I did gives me a pace that I can fully control. TDD helps me avoid infinite loops of confusion and frustration.

The more defined development cycle also reminds me to review my code often. The TDD cycle offers a natural moment to reflect: as soon as the test passes. When all my tests are green, I consider whether there is anything to improve in the current code.

Designing classes is one of the most challenging tasks of a software engineer. I appreciate the TDD cycle because it forces me to use the code I am developing from the very beginning. The perception I have about the class I am designing is often different from my perception when I try to use the class. I can combine both of these perceptions and make the best decision about how to model the class.

If you write the tests after the code, and not before, as in TDD, the challenge is making sure the time between writing code and testing is small enough to provide developers with timely feedback. Don’t write code for an entire day and then start testing—that may be too late.


Comments

Leave a Reply

Your email address will not be published. Required fields are marked *