I'm building an installable python package, it's basically a CLI based on docopt and other herbs.
The overall structure is as follows:
README.md
MANIFEST.in
setup.py
grok/
__init__.py
module.py
other_module.py
main.py
On my main.py I have the other modules imported as follows:
main.py
# Imagine the docopt USAGE string here
from .module import something
from .other_module import other_something
# the rest of the logic comes here
The main issue is, I can install the package and use it, but, when I'm changing the code, maybe adding features or fixing a bug, if I try to run it like this:
python main.py -h
It gives me the following error:
ModuleNotFoundError: No module named '__main__.module'; '__main__' is not a package
I googles a bit and found out that because of the dot on my imports I'm actually referencing __main__, so, I should remove the dots and I should be able to test it the way I intended after that.
I did so, and it worked, I tested all my stuff, and then ran the test suite, as a part of the suite it tests installing the package and run it with -h as a way to smoke-test it, and it failed with the following error:
ModuleNotFoundError: No module named 'module'
Then I re-added the dots to the imports and the test passed.
What am I doing wrong? in theory I should be able to import everything in a way I can execute the installed package and run it on my code for testing without having to change anything.
Any hints?
