TypeVar - preserving type identity

Question

Accepted Answer

The problem TypeVar solves is expressing 'the output type depends on the input type' - what type theory calls parametric polymorphism. If you write 'def first(items: list[Any]) -> Any', a type checker knows nothing about what first() returns. Callers lose all type safety after the call: mypy can't know that first(['a', 'b']) is a str. With 'T = TypeVar("T"); def first(items: list[T]) -> T', you tell the checker: 'whatever element type the list has, that's what comes back'. Now mypy knows first(['a', 'b']) is str and first([1, 2]) is int, and it will flag downstream mistakes accordingly. TypeVar does nothing at runtime - T is just a placeholder that the type checker uses to track the relationship between argument and return types. You can also bound a TypeVar: 'T = TypeVar("T", bound=Comparable)' restricts T to subtypes of Comparable. For data scientists, this matters most when writing generic utility functions that work on DataFrames of different dtypes or arrays of different shapes.