I’m writing a a bit about pipeline inputs and outputs in all of the

README.md

files of my project. How should I document the

type

of a partitioned dataset? Those function outputs need to be of the form

dict[str, <type>]

, but when the dataset is loaded back it’s going to be

dict[str, Callable[[], <type>]]

Nice, thanks!

if you’re on Python 3.10+ there is a slightly nicer syntax too

I am on 3.10, I can use lower case

dict

now, right?

you don’t need union anymore

Ooh, nice. I like this PEP

💪

Thanks again!

np

dict

works without any extra imports since 3.9; you can also do

from __future__ import annotations

for past versions.

Callable

, if used, should have 0 or 2 arguments (list of inputs, output).