TypeScript'te tip çıkarımı nasıl çalışıyor ve ne zaman manuel tip tanımlamak gerekir?

02 Kas 21:49

TypeScript’s inference is pretty solid for straight‑line code—assignments, literal returns, and simple callbacks are inferred the same way a compiler for a statically‑typed language like C# would deduce the type from the right‑hand side. In practice it feels as reliable as Flow’s inference, but unlike Flow you get a concrete type (e.g., `string[]`) baked into the emitted declaration files, which helps downstream consumers. Where it starts to wobble is when you bring in data from the outside world (JSON from an API, `any` from a third‑party lib) or when you nest generics heavily (e.g., `Promise<Result<T>>`). In those cases the compiler can only infer `unknown` or a very broad union, so you lose the safety net. My rule of thumb: add explicit annotations whenever the inferred type is wider than what you actually expect, especially on public interfaces, function signatures that are part of a library, and any data that crosses the network boundary. Think of it like the difference between letting Go infer the type of a struct field versus tagging it with a JSON schema—explicit tags prevent mismatches later on. So for internal helper functions you can rely on inference, but for API contracts, generic utilities, and anything that other modules import, write the types out front. This keeps the codebase predictable and makes IDE tooling as helpful as it is in a language like Java.