Class implementations in __init__.py

Question

I have been looking into __init__.py files, and see the recommendation is to handle the imports for the package there, and/or some initialization.

Then there is a package like Python Collections. That __init__.py contains the implementations for different Collection objects.

Why would you place implementations in the __init__.py instead of dedicated .py files within the package? Is this a good idea? Is it a design pattern?

Answer 1

In this particular case, there's no need to have a directory with an __init__.py file at all, since the only other file in that module is abc.py, which defines nothing itself. So if it weren't for abc.py, the entire thing should just be collections.py instead of collections/__init__.py. So why is it what it is? Purely for backwards compatibility reasons. They did not want to break from collections.abc import ..., but wanted to move stuff around internally.

• sanyassh: Do you have any thoughts in general, not about this particular case of collections?
• deceze: It's generally not a terrible idea IMO. You only want to be careful with circular imports, e.g. when a sub-module imports some base classes from __init__.py, but __init__.py also imports those submodules to expose them.