Code reuse considered harmful

The title is intended to call for attention. This post is about one perspective of software development in the light of my own experience in the area, it won’t contain anything really revealing and is not to be taken as an absolute true for life. It’s a rant. I hope you have a good time reading it, feel free to leave me any kind of feedback.

I see a bunch of people praising reuse as being the prime thing of good software development, and few talking about replaceability. There seems to be a constant seek to avoid writing code that is used only once, as if it was a really bad thing. Then we end up with software that is made of conceptual factories that create factories that create the things the software really needs, yes there are two levels of factories, or more. Is this really necessary? How much time do we save by this extreme look for reusing code?

First, let me ask and answer a simple question: why duplicated code is annoying? Well, duplicated code makes it harder to change stuff. When you have the same piece of code written multiple times in a code base and you find that it needs a change, e.g. bug fix or new feature, you will need to change it in all places. Things can get worst if you don’t know all places where the code is duplicated, so you may forget to change one of these spots. The result is that duplicated code is a sign of harder maintenance and a fertile ground for further bugs to spawn. That’s why we learned to hate it. We started fighting this anti-pattern with all strength we had.

Code reuse is the perfect counter to code duplication, right? Sure, it is right, if we reuse a piece of code in two places, we have no duplication between these places. So, we did it! We found the Holy Grail of code quality, no more duplicated code, yay! But something unintended happened. Remember the old saying: with great powers, comes great responsibility. People started to be obsessed with it. As soon as they learned to use the hammer of code reuse, everything turned into a nail, when it didn’t work out in the first hit, they adjust the size of the hammer and hit it again with more effort.

This seek after code reuse led us to a plethora of abstractions that seems to handle every problem by reusing some code. Don’t get me wrong, lots of them are useful, these are the ones that were created from observation. The problem is the ones that are created from “it’s cool to abstract”, or other random reason that is not true observation. We see frameworks after frameworks that try to fit every problem of the world into a single model. Developers learn to use these frameworks and suddenly find out that the framework creator is wrong and create yet another abstraction over it or creates yet another framework that tries to use a different model to solve the world.

What happens when we have a bug in one of these abstractions or we need to enhance it? Silence, for a while, then the sky turns black, you take a break, go for a walk, come back to your computer and start blaming the other developer that created the bug or that “got the abstraction wrong”, because your vision was the right one. What happened? We reused code to avoid code duplication, but we are still having the same problems: code that is hard to maintain and evolve.

My guess? We missed the enemy. Code duplication is not our enemy. Maintenance problem and rigidity of code is.

My tip? Give more focus on replaceability of code instead of reuse in your talks, codes, classes, etc. Create the right abstraction to fix the problem at hand in a way that is easy to replace the underlying code when needed. Some time in the future, you will need to change it anyway. That’s what agile methodologies try to teach us: embrace change. Planning for a design to be reused says: “my design will be so awesome, that I will reuse it everywhere.” That’s what agile says: “your design will need to change sometime, because the requirements will change, plan for the replaceability of it.” People are doing things like service oriented architecture in the wrong way because they are looking for reuse of services and not for replaceability of services, they end up with a Big Web of Mud.

That’s all folks. Thanks for your time.

Advertisements

One thought on “Code reuse considered harmful

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s