TDD — Lookahead Refactoring

Image by pixel2013 via Pixabay

In this article, I want to consider a situation where considering the next test we want to write, influences the way we refactor.

Aside

This article is the third one I’ve written recently looking at interesting little nuances in the TDD process (see TDD — Choosing the Right Intermediate Steps and Which Order to Write Your Tests). It was not my intention to create this mini-series on the topic, but it has been an exciting outcome of working on creating training material to teach TDD from the ground up. During this work, I have been learning these delightful new things — not in complex code, but in the simplest of problems possible.

I consider myself a somewhat seasoned TDD practitioner, but through discovering new things in the absolute basics (where it would seem that there is nothing left to learn) highlights the importance of continually reviewing the foundations of our skills.

For this article, I want to consider implementing a simple function to check if a number is odd or even. The function will simply return true if it’s given an even number, and false if it’s given an odd number.

As with my previous TDD articles, we are working very rigidly with the TDD cycle — we must write a failing test, we must implement it in the simplest way possible, we may then refactor.

The Example

For this example, let’s ignore the fact that Ruby’s built-in Integer#even? method exists.

Step 1

We start with the following test:

it 'returns false for 1' do
  expect(even?(1)).to be(false)
end

We make it pass by simply returning false:

def even?(number)
  false
end

Step 2

We add the next failing test:

it 'returns true for 2' do
  expect(even?(2)).to be(true)
end

We make it pass by adding a guard clause:

def even?(number)
  return true if number == 2

  false
end

And then we refactor:

def even?(number)
  number == 2
end

Step 3

Now things get interesting — I’ve been working up through the numbers, but writing a test for 3 (which would be the next logical progression) will not fail. I want to stick very rigidly to the TDD cycle, so wehave two options:

1. Choose an even number for my next test
2. Refactor to allow us to write the next failing test

The first option might seem like the obvious one, but this article is about investigating the second one. Before writing the next failing test, let’s refactor again:

def even?(number)
  number != 1
end

At this point, the tests still pass. Now we can write the next failing test:

it 'returns false for 3' do
  expect(even?(3)).to be(false)
end

Once we have seen it fail, we can write the code to make it green:

def even?(number)
  number != 1 && number != 3
end

Now it’s time to refactor. We could implement our final solution now but I don’t like that idea. We currently have a lack of symmetry, we have two tests for the odd result but only one for the even — that bothers me. I want to add another test for the even case but I can’t because we can’t write one which will fail (I want to see it fail because it proves I’ve written the right test). So what can we do?

Step 4

Let’s refactor again:

def even?(number)
  number == 2
end

Again, the current tests pass, so now we can write the next failing test:

it 'returns true for 4' do
  expect(even?(4)).to be(true)
end

Finally, we can complete the implementation:

def even?(number)
  number % 2 == 0
end

That’s not refactoring!

Technically some of the refactorings I did above are not refactorings; they change the outward behaviour of the function. However, in the context of tested behaviour, they are refactorings.

What I want to point out here is that when you use TDD to create new code, there is a different state of mind — we know we are in the process of creating something new, and that we are not finished yet.

Note: I have since followed this up with a second article which looks at another way to achieve this while sticking strongly to the definition of refactoring.

Why?

What I have tried to highlight here is that often when doing TDD we’re not just trying to implement the final solution, but we’re also trying to create a test suite which gives us confidence.

Sometimes we can do TDD absolutely by the book and arrive at the solution we want, but feel we are missing some tests. In this case, the temptation is often to just add the additional tests, but when we do this we never see them fail. For me, I’m almost more concerned with seeing a test fail than seeing it pass because I want to be confident that I haven’t made an error in the test code.

when we are practising TDD we must keep thinking ahead about the solution we are designing towards, even though were are focusing on each individual step along the way.

What this example shows, is that during our refactoring stage, it can help to consider what we are trying to do next. This ties in with the same message I was trying to highlight in my previous articles — when we are practising TDD, we must keep thinking ahead about the solution we are designing towards, even though we are focusing on each individual step along the way.

I hope you found this little thought exercise interesting.