I assume you meant using leetcode's editor without type checking then yeah, there's much less benefit using type hints there.
In my vscode setup with mypy strict, pylance basic and python 3.9 though, type hints all days for me. I personally forget the arguments types like 1 second after I write the function. Maybe I'm conditioned by typed python.
Leetcode codes usually have simpler data structure and without 3rd party libraries so that's even easier to add types for.
Edit: Oh looks like Leetcode python code template has type hints
I would argue it's less readable, substantially. Type hinting was never about readability, it was about getting IDEs/tools to do type checking for you.
def max_width_of_vertical_area(points):
xs = sorted(x for x, _ in points)
return max(x2-x1 for x1, x2 in zip(xs, xs[1:]))
Although I see both sides, I personally lean towards finding that type hints improve readability. And since I tend towards consistency, that ends with most/all functions being typed.
Though I spend most my time writing Swift code, which definitely influences my preference.
And then who-ever reads your code has no idea what "points" is. We had readable "type-annotation" 10+ years ago with comments to functions, this is just a logical extension to that. People need to know how to call your function, whether it's a statically-typed language or not.
Would you say this is "less readable substantially" than your example?
def max_width_of_vertical_area(points):
# type points: List of List of int
xs = sorted(x for x, _ in points)
return max(x2-x1 for x1, x2 in zip(xs, xs[1:]))
I would never comment that such a function takes a `List of List of Int` - I would instead describe `points` as being something like an `Iterable of Pairs of Numbers`.
Are Python type hints able to specify types like that?
The new type-hints sure: List[List[int]] (and you can do this to any arbitrary types or levels of nesting. See fancy example at the end.). But older iterations before the formal type-hint spec had varied/non-standard ways of doing it. Some machine-readable but others more for the user (like List of List of int). Even once the type-hint spec got approved, it was backported to older versions of python using docstrings.
My experience even before that was mostly with PyDev in eclipse. The trick back then was identifying good spots to add type-hints, afterwhich the "IDE" would take over and propagate them through the different variables and calls using static code analysis. It was surprisingly good and helped immensely with very few actual instances of types being hinted. Another one was PyCharm that would "save" or "cache" the types during debug/test runs and use them as type-hints without you explicitly having to define them.
Fancy example:
List[Dict[str, MyCustomTypeGeneric[str]]]
Which gives you a list of dictionaries with keys of type string, holding values MyCustomTypeGeneric with generic type str.
leetcode adds them automatically and I just didn't bother to take them off. leetcode also names the functions with camelCase versus whatever_this_is_called (which is the python way).
