Last week my colleagues at Gutefrage.net and I had our second
Coding Dojo.

I learned a important lessons - even though this wasn’t the first time I practiced the kata.

The kata

The task was to create a class that decomposes natural numbers into its
prime factors.

Additionally the factors should be sorted ascending.

Examples:

You get the idea ;)

This sounds simple and like a perfect candidate for an
introduction to TDD right?

Unfortunately this kata is a bit tricky.

Performing the kata

The first tests are straightforward:

Now the trouble begins

Now decomposing 5 would be a nice test? Right? Since 5 is a prime it
would only return 5. But how can we find out if its a prime?

So we could write a function that finds out if 5 is prime. Hm that seems
hard.

What about 6? But then we have to develop a algorithm that works -
otherwise we would just move sidewards and add conditionals.

Whats the problem?

You do not know how to solve the problem. Not everything is as simple as
the FizzBuzz Kata.

Adding another test alone does not suffice. It’s a common hurdle for
newcomers to TDD.

How to solve this?

Grab a sheet of paper and think about the algorithm. Or do a little spike until you fully understand the
problem. Sometimes it’s just not true, that a design “magically” emerges.

Learnings

You need a sound knowledge about where to go - otherwise you will have problems with TDD along the way. Then you have to lean back and think a little bit or get feedback from something else then your tests.

TDD is cool, but it can not solve every problem. Especially not complicated algorithms (this is a really easy one of course).

Another personal learning for me was to watch myself better, when I do a kata the first time.

When the design does not emerge from the tests - I should listen to them. Maybe they want to tell me that I have absolutely no idea what I am doing and that I need to think more..