Python 3.x introduced a relatively unknown feature called “function annotations”. This introduces a way to tag parameters and your return value with arbitrary information at the function-definition level.

You can annotate using strings or any other type that you’d like:

>>> def some_function(parm1: "Example parameter"):
...   pass
... 
>>> some_function.__annotations__
{'parm1': 'Example parameter'}
>>> x = 5
>>> def some_function_2(parm1: x * 20):
...   pass
... 
>>> some_function_2.__annotations__
{'parm1': 100}

You can also annotate the return:

>>> def some_function_3() -> 'return-value tag':
...   pass
... 
>>> some_function_3.__annotations__
{'return': 'return-value tag'}

It’s important to note that there are already strong conventions in how to document your parameters, thanks to Sphinx. Therefore, the utility of annotations will most likely be entirely in terms of functionality. For example, you can annotate closures on-the-fly:

import random

c_list = []
for i in range(10):
    def closure_() -> random.random():
        pass

    c_list.append(closure_)

list(map(lambda x: print(x.__annotations__), c_list))

This is the output:

{'return': 0.9644971188983055}
{'return': 0.8639746158842893}
{'return': 0.18610468531065305}
{'return': 0.8528801446167985}
{'return': 0.3022338513329076}
{'return': 0.6455491244718428}
{'return': 0.09106740460937834}
{'return': 0.16987808849543917}
{'return': 0.9136478506241527}
{'return': 0.41691681086623544}

Absolutely nothing in the Python language or library is dependent on annotations, so they’re yours to play with or implement as you see fit.