Python range is too narrow

[EricKolibacz]

Description

When trying to add maxi-lib to a project, and the project is simply set up with python 3.8, poetry will raise an error. Related to #16

Steps to reproduce the behavior

Create a repo and initialize with poetry

poetry init

(and accept default, without adding main and development dependencies)

And then add maxi-lib:

poetry add git+https://github.com/dailab/MAXi-XAI-lib.git

You should get following error:

The current project's Python requirement (>=3.8,<4.0) is not compatible with some of the required packages Python requirement:
  - maxi requires Python >=3.8.0,<3.10, so it will not be satisfied for Python >=3.10,<4.0

Because <repo-name> depends on maxi (0.4.2) @ git+https://github.com/dailab/MAXi-XAI-lib.git@HEAD which requires Python >=3.8.0,<3.10, version solving failed.

  • Check your dependencies Python requirement: The Python requirement can be specified via the `python` or `markers` properties
    
    For maxi, a possible solution would be to set the `python` property to ">=3.8,<3.10"

Minimum Acceptance Criteria

• When following the steps above, poetry should not raise an error when adding maxi-lib

[EricKolibacz]

Interestingly, the suggested solution does not seem to be correct. Two options are probably available:

• set the python property to ">=3.8,<3.10" but for the repo
• set the python property of maxi to python = "^3.8"

The second solution should be preferred because the first solution requires intervention of the user. Hence, solution 2 is more user-friendly.

One counter-argument against solution 2:

• some python 3.8 specific language is used, e.g., type hint format.

Python 3.10 and 3.11 are downwards compatible, if I am not mistaken. So this shouldn't cause any error.

[EricKolibacz]

Tensorflow limits the python version range which causes the version incompatibility of maxi-lib with any freshly initialized poetry project. No solution found on the internet, that's why I posted a question on stackOverflow.

[EricKolibacz]

Let's wait for answer in StackOverflow. Otherwise, the current solution from #18 is most probably the best.