Priority of .egg-info, .dist-info, and setup.py

[bmw]

• I am on the latest Poetry version.

• I have searched the issues of this repo and believe that this is not a duplicate.

• If an exception occurs when executing a command, I executed it again in debug mode (-vvv option).

• OS version and name: macOS 11.3.1

• Poetry version: 1.1.6

• Link of a Gist with the contents of your pyproject.toml file: https://gist.github.com/4df62bef5bccf68667b8dd31c62e2a90

Issue

This is arguably not a poetry issue, but I think it'd be nice if poetry or its dependencies did something to avoid the problem I'm having. In addition to the pyproject.toml file I linked above, my toy example here also contains the directory foo whose only contents is this setup.py file.

To reproduce the problem I'm having, do the following:

1. Run pip install -e foo.
2. Modify the setup.py file to depend on another package such as requests.
3. Run poetry lock.

The problem here is that step (1) creates a foo.egg-info directory which poetry lock uses in step (3) despite the setup.py file being modified to include other dependencies in step (2). Using the packages from my example here, poetry.lock only includes urllib3 and not requests which I added as a dependency in step (2).

Can poetry or its dependencies do something to avoid this issue?

[dimbleby]

probably duplicate #3783, cf #6843

[radoering]

Actually a different issue than #3783 which is about poetry lock --no-update. Here, poetry lock is also affected.

Interestingly enough the following works:

1. Run poetry lock
2. Modify the setup.py file to depend on another package such as requests.
3. Run poetry lock (lock file is updated correctly)

but the following doesn't:

1. Run poetry lock
2. Run poetry install (egg-info is created)
3. Modify the setup.py file to depend on another package such as requests.
4. Run poetry lock (lock file is not updated)

[finswimmer]

At the moment .egg and .dist-info folders are parsed for metadata before trying to parse setup.py:

poetry/src/poetry/inspection/info.py

Line 490 in 2a70d67

info = cls.from_metadata(path)

So the only way to solve this, would be to change the order. No idea if this would have any negative site effects 🤷‍♂️