Why Didn't You Use a Matrix There?

As we were writing tie, Mark says "I think we need a Matrix about now."

Now I know I'll eventually need a matrix (or 2 dimensional array, or something like that) eventually. But nothing in our tests requires that we use a matrix yet. If I can satisfy the tests without using a matrix, then I will do it.

This is called Doing the Simpliest Thing That Will Work, and is one of the slogans of Extreme Programming. There are good things that come from following this slogan.

• It keeps the code simple (see What is Simple). I love complex things and one of the things I had to learn as I starting XP was to keep things simple. Implementing only what's called for by the tests keeps me honest.

• By keeping the implementation simple, I concentrate on the interface first. I review the tests and ask the question "Does this way of using the class satisfy my needs?" Once I'm happy there, I can move on to more complete implementations.

• When I am able to write code that I know is too simple (but still passes the current set of tests) then I know that my tests are not complete. So I ask myself: "What test can I write that will force me to generate the code that is needed?"

So let's ask ourselves that question ... What test can we write to force us to use a matrix?

[ next ] [ prev ] [ contents ] [ skip to Testing in a Matrix ] [ up to Test Number Three -- Tying the Knot ]

Copyright 2003 by Jim Weirich.