Are there advantages of Python3 type hints over Python2?

Question

As I understand it, Python2 had a type hint convention which resembled:

def geo_series(r, n) # type: (float, int) -> float
  return (1-r**n)/(1-r)

This has been replaced with Python3's system which looks like:

def geo_series(r: float, n: int) -> float:
  return (1-r**n)/(1-r)

My question is: what is the advantage of the Python3 system over that of Python2? It seems to me that they contain equivalent information. Since the type-checking is done by an external tool anyway, I see no major disadvantage to putting the type-hint in a comment.

Am I missing something? My best guess is that there are situations in which you can create some wacky type which makes the python3 type-checking more informative than the equivalent python2?

Answer 1

As Patrick mentioned in the comments, type annotations are part of the code, which means they can be accessed programmatically. For instance, in your python 3 example, you can use

>>> geo_series.__annotations__
{'r': float, 'n': int, 'return': float}

This programmatic access makes it possible to dynamically type check functions if you want to implement that.

An example of annotations being used within code is the dataclasses module . A field in a dataclass is detected based on whether it has a type annotation, which is only possible if the annotations are part of the code instead of in a comment.