I was recently watching a talk by Dan North. At one point he briefly talks about a project he worked on. For this project, the Ant build script was generated with an XSLT transform, producing a build script about 2 million lines long. Even though that is clearly horrible, my first fleeting thought was, Huh, I would have never thought to do that. That's pretty clever..
Dan North's Simplicity, the Way of the Unusual Architect, at the time of the story
That's pretty clever, is absolutely the wrong thought to have. I cannot fathom the monumental, wasted effort maintaining such a system. But I am also sure it started as someone's clever idea. There would be plenty out there who would say, Well, that project went wrong, but the idea itself isn't bad. It's pretty darn clever, really.
"Cleverness" like that always seems promising at first. It is a novel view of a problem at hand. Because clever ideas are novel in completely unexpected ways, they have unexpected consequences. Once in place, they become entrenched in the structure. Just implementing those ideas was an undertaking because of their cleverness. Throwing that all away is not an option. Keeping them working under evolving circumstances requires layering on additional cleverness. It all looks like a precarious game of Jenga. There are quirks to build around. There are blocks never to be touched.
Brian Kernighan said, Everyone knows that debugging is twice as hard as writing a program in the first place. So if you're as clever as you can be when you write it, how will you ever debug it? Those of us who are not clever enough are trapped. We plod through the complexity just trying not to break anything. Most importantly, we will all end up not being clever enough eventually.
We all seem to admire cleverness, despite its dark side. From a young age, we are praised for our cleverness. It becomes an admirable trait. Cleverness "on the computer" might even be why we do what we do. That's why we momentarily admire horrifying things, like XSLT transforming Ant scripts. Cleverness is not intelligence, though. We can be smart without doing clever things.
Smart people like to be challenged. Being clever is one sort of challenge, but there are others. A better challenge is to be dull. Think of the most boring, straightforward way of doing something. Make it just flexible enough, but don't over engineer. If it is simple enough for a brand new software developer to understand, you've achieved your goal. When you produce software that dull, you produce software that is maintainable, software that is able to evolve.
As software developers, we see errors every day. They manifest as exceptions or segfaults or error codes, telling us that our code has gotten into a state we didn't expect. Their appearance often portends bugs. Though we groan at an unexpected stack trace, we should see an error as a form of automated feedback . Feedback can be fast or slow. We can put the discovery of errors onto a timeline. Errors appear at many points over the lifetime of the code, starting at the moment it is compiled. The further to the right that an error appears, the longer it takes for the feedback to appear. Compile time Compile time is the earliest we can receive feedback about an error. The compiler automatically does a number of checks to make sure the code makes some semblance of sense. The most powerful tool for compile-time feedback is the type-checker. The type-checker makes sure that only values of the expected type are passed around to the places that expect them. This guards
Magic numbers are considered such a scourge upon code bases that there tools out there to automatically find them and root them out. After all, who knows what setValue(6); really means? There are also many approaches to giving better names to magic numbers. Not all of them are good, though. For example, we have a callback that sets a completion percentage on a status: void callback(Status status) { status.updateProgress(0, "Starting"); // Do some stuff... status.updateProgress(25, "Some stuff done"); // Do some more stuff status.updateProgress(50, "More stuff done"); // Reticulating splines status.updateProgress(75, "Splines reticulated"); // Validate everything status.updateProgress(100, "Complete"); } Those percentages would get flagged as magic numbers by the automatic tools. The developer would need to extract those out to constants to make that tool happy. A thoughtful developer would create
Sometimes when developing code, I run into problems. Ok, a lot of the times I run into problems, but there's a certain kind of problem I run into pretty often. The problem is that I want a thing that doesn't exist. I am writing code, and it is just getting gnarlier and more tangled around a pile of logic. At some point, I hit a wall with it. It is too complicated to really reason about, and I don't want to make things any worse. I've found a trick to solving these problems, though. I take those problems and give them a name. Once I do that, I can get a handle on them, manipulate them, or even sequester them off into a corner. In almost every case, this drastically reduces the complexity back down to a reasonable level, and I often get the side benefit of offering extensibility I didn't know I needed. To demonstrate, let's say we are working on an e-commerce site. The company is based in Alabama, and sells to customers in all 50 states in the US. We have a met
Comments
Post a Comment