The Methodical Margin
Sign in

Writing

Posted 5 minutes ago · 30,403 reads

Legacy code is called legacy for a reason. It works. It's been tested in production. It's often the most reliable code in your system, even if it doesn't follow modern conventions.

APIs are contracts. Once you publish one, changing it becomes expensive for everyone who depends on it. The cost of breaking changes compounds over time, which is why the boring, conservative choice is usually the right one.

The best code is often the code you don't write.

The most important insight I've had in the last few years is that constraints are a feature, not a bug. When you have unlimited resources, you can solve any problem in a hundred different ways. When you have constraints—limited memory, limited time, limited developers—you're forced to think more clearly.

The most important insight I've had in the last few years is that constraints are a feature, not a bug. When you have unlimited resources, you can solve any problem in a hundred different ways. When you have constraints—limited memory, limited time, limited developers—you're forced to think more clearly.

Most of the code we write is not rocket science. It's ordinary business logic, wrapped in layers of frameworks and abstractions. Sometimes the simplest implementation is the best.

Most of the code we write is not rocket science. It's ordinary business logic, wrapped in layers of frameworks and abstractions. Sometimes the simplest implementation is the best.

Related

More on this topic

I've noticed that the best technical decisions come from understanding not just what works, but why it works. The deeper your intuition about a system, the better your architectural choices become.

Understanding the problem is half the solution.

Terminal emulators are primitive tools compared to modern IDEs, but they force a certain discipline. You can't rely on auto-completion and syntax highlighting to write code for you; you have to understand what you're writing.