30 Jun 2014 Think a Bit, Code a Bit, Test a Bit
The best piece of advice I ever got regarding a personal software development process was from a grizzled old Unix developer with a neckbeard. OK, that’s not true – it was actually from a clean-shaven principal consultant and architect at a company I used to work at, but that doesn’t sound nearly as impressive.
Nevertheless, the process went something like this:
- Think a bit
- Code a bit
- Test a bit
- Go to Step 1
Having tried all manner of processes over the years, this is the one that has served me best. Let me break it down for you if you’re having trouble understanding it.
Step 1: Think a bit
Do you equate productivity with the sound of keys clicking on a keyboard? If so, my advice is to slowly step away from the computer and actually have a think about the problem you’re trying to solve before leaping into the fray.
When it comes to thinking, some people have trouble knowing where to start. If that’s you, here’s a suggested starting point: what are you actually supposed to be doing? Sometimes a few moments of quiet reflection will reveal that what you were thinking of typing doesn’t really relate to what is actually needed, or that you need more information before you can continue.
The next thing I recommend thinking about is: at this point in time, what’s The Simplest Thing That Will Possibly Work? I have never regretted asking myself this question; I regularly regret not asking it before I started typing. Also known as Keep It Simple, Stupid (KISS) or You Ain’t Going to Need It (YAGNI), the quest for The Simplest Thing That Will Possibly Work is a noble and worthy goal for a software developer.
Feel that The Simplest Thing That Will Possibly Work is insufficiently challenging for your superior skills? Maybe, maybe not. Sometimes, when you actually force yourself to think about it, finding and executing The Simplest Thing That Can Possibly Work can actually be quite a challenge in itself.
Complexity will come and find you eventually; there’s no need to seek it out prematurely. Besides, what’s the worst that can happen when you do The Simplest Thing That Will Possibly Work? You can always just go to the pub if you finish early (although funnily enough, this has never happened to me, ever).
That said, you can only do so much thinking about the problem you’re trying to solve and how you’re going to solve it. It’s important not to go overboard and enter a state of analysis paralysis – a debilitating condition I am personally prone to. My strategy: as soon as I feel the symptoms of paralysis setting in, I know that it’s time to move on to the next step. Which brings us to…
Step 2: Code a bit
Now for the fun part. I’m not even really going to go into this one too much – it’s why we’re all in the computer programming game, right?
But of course, one can have too much of a good thing and coding is no exception. Soon unwanted thoughts and questions will begin to enter your head – for example: how do you know if your program is going to…you know…DO WHAT IT’S SUPPOSED TO DO?
So let’s now artfully jump-cut past your ecstatic communion with the keyboard and go straight to the next step, when it’s time to…
Step 3: Test a bit
I’m not going to lie to you and pretend that this is as much fun as coding. That said, this step isn’t really negotiable.
“But Ben”, I hear you say, “I don’t want to do stuff that’s boring”. To you, I say: too bad, you’re not being paid to produce code that doesn’t work. Never in the history of computing has it been easier to write automated tests. So stop procrastinating and write some tests already. If you really want to write code without tests, do it in your own time, on your own open source project (which no-one will use, by the way, because it has no tests).
And perhaps, to be honest, our aversion to testing stems not so much from the fact that it’s boring, but from the fact that’s it can actually be quite hard. In much of the work I do these days, testing is probably the most challenging part because, unlike the coding phase, the optimal solution is less obvious.
Writing tests that are both meaningful (i.e., not just mocking the heck out of everything) yet practical (i.e., not excessively fragile or long-running) is hard. I’d call it more art than science, and the ideal solution differs from project to project. Unlike the noble quest for the Simplest Thing That’s Going to Work when coding, testing can be a messy, uncertain affair.
Some people like to do this step before they code. I say: good for you; it’s nice if you can do it, but don’t get hung up on it. If what I’m doing is highly algorithmic, test-first can be useful. But that doesn’t happen to me often – most of the work that I do is essentially integrating existing pieces. And sometimes I don’t know exactly how to do that until it’s been done, which kind of makes TDD hard.
That said, whether you test-first or test-after, you should always test.
Step 4: Rinse and Repeat
We can’t expect to solve all of a problem in one hit; iteration and feedback are critical. Developing a piece of software is a process of continuous learning – about both the technology and the problem domain. A tight cycle of planning, execution and reflection is the optimal strategy.
As we progress, this cycle may accelerate, sometimes to the point where the steps will appear to blur into one another. But if we hit a roadblock and progress falters, this does not mean that the steps should be skipped – we should just slow things down and continue our cycle of Think A Bit, Code A Bit, Test A Bit.
So there you have it, a simple 4-step personal software development process. It’s served me well for most of my career, and the good news is that modern software development tools are making it easier and easier to work this way. Furthermore, whilst this approach predates the agile movement, it obviously ties in with the agile philosophy.
So try thinking a bit, coding a bit, and testing a bit – I promise that you won’t be disappointed.