Lessons from the prime factorization dojo

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

Coding Dojo.

I learned an important lesson — 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:

1  => [] # Since 1 is no prime ;)
5  => [5]
10 => [2,5]
75 => [3,5,5]

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:

module PrimeFactorDecomposer
  def decompose(number)
if number < 4
    number
else
    [2,2]
end
  end
end

describe PrimeFactorDecomposer do
  include PrimeFactorDecomposer

  it "returns no prime factors for 1" do
    decompose(1).should be_empty
  end

  it "decomposes 2 into 2" do
    decompose(2).should be == [2]
  end

  it "decomposes 3 into 3" do
    decompose(3).should be == [3]
  end

  it "decomposes 4 into 2,2" do
    decompose(4).should be == [2,2]
  end
end

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 it’s 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 an algorithm that works — otherwise, we would just move sidewards and add conditionals.

What’s 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. Every so often it’s just not true, that a design “magically” emerges.

Learnings

You need 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 or get feedback from something else than your tests.

TDD is cool, but it cannot solve every problem. Especially not complicated algorithms (this is a straightforward 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…