Effective TypeScript by Dan Vanderkam: Key Takeaways

In this post I'll summarize some of the high-level advice from Dan Vanderkam's book on how to build a mental model of TypeScript. Instead of diving into code and inner workings, I'll focus on what I found to be the most helpful ways to view how the language works.

1. You should think of types as simply a set of possible values. Any check in TS is basically checking if one type is a subset of another. There's a reason type names are called type aliases; the name is unimportant, and TS is simply comparing the contents to something else.
2. Let TS infer types. TS is pretty good at guessing the types of values, and adding explicit types can clutter code and cause issues if not updated when code is. Cases to use explicit types are:
   • TS doesn't know what the type is (e.g. data from an API)
   • Return types on functions - though TS can often infer here, explicit return types can prevent mistakes
   • Object literals, to make use of excess property checking
3. Stricter is pretty much always better. More strictness means more ways bugs can be caught at runtime. An obvious example here is avoiding any, but some others are:
   • Ensure tsconfig is strict. Two important flags here are strictNullChecks and noImplicitAny
   • Only use type assertions when TS can't know what the data type is
   • Use readonly when applicable
4. DRY applies to TS. If you wouldn't duplicate it in code, you shouldn't duplicate it in types. For all the same reasons!