I was looking at a great talk on Functional TypeScript by Thiago Temple in which he goes over a use-case (and general reason) for not using null whenever possible, but instead enforcing stricter type-checking to cater for empty return types in a homogenous manner.

A value of null can have any number of meanings, including but not limited to:

• Value not found
• An error has occurred
• Unpredicted case
• Value not required
• Value is falsy

null therefore hides contextual meaning whenever it’s used. Consider the following function:

function findUserById(id: string): User | null {
  // ...
}

If null is returned, then it could mean either of the following:

• the User could not be found
• an Error has occurred

Since null doesn’t provide clarity, creating an Option type that unions a generic Some type and None type can help to give meaning to the output:

type Option<T> = Some<T> | None

type Some<T> = {
  kind: "someOption"
  value: T
}

type None = {
  kind: "noneOption"
}

With these types crated, the findUserById function’s output will indicate that either one of two options are present in the response; a clear value (of type Some) or no value (of type None):

function findUserById(id: string): Option<User> {
  // ...
}

Because of the kind property is shared between the two types, it can be inspected and used for conditional logic:

const user = findUserById;

if (user.kind === "someOption") {
  user.value;
  // ...
}

Links

• TypeScript Talk by Thiago Temple - https://www.youtube.com/watch?v=fhyHgkH0ZEg