Add automated tests and CI

[smithara]

Following on from #11 (comment)

Since there are more users of chaosmagpy, and some history of Python versions accruing and some installation issues, I think it's time to set up automated testing. I'm working on this here: https://github.com/smithara/ChaosMagPy/tree/setup-ci

Are there already some thoughts/plans here? I don't want to step on any toes :)

At first glance, I would:

• test against the main recent Python versions. There's some balance to find about how far back to test - it's fine to still allow installation on old versions (where a user might need to make some manual intervention to make the install work), but we don't want to bother testing and maintaining too far back
• Run the test weekly (is this a reasonable rate?) so that we can catch installation issues when dependencies are updated
• need to adjust the current tests to supply the test data directory
• look at removing the test_package.sh, build_package.sh, and consider reworking to use pytest

[ancklo]

Hello,

Yes, I think this is a good idea. This was always in the back of my head, but I never got around to looking into this. So any help on this is appreciated :)

I made the script test_package.sh when I ported our in-house MATLAB functions and scripts over to Python. I'm sure we can reuse some of those tests.

[ancklo]

I merged #14 into the main branch and added a trigger on push to master.

I'm currently preparing a new release of ChaosMagPy (v0.15), which includes a method to evaluate the ionospheric field in CHAOS-8. This change will make ApexPy another dependency, which is sometimes difficult to install due to the underlying Fortran. However, I only plan to make this feature optional at this point.

According to the documentation, apexpy should work with Python 3.12, but I only managed to get it to work on Python 3.11, at least on my computer.

Would you recommend that I add the following to the optional dependencies in the pyproject.toml?
"apexpy>=2.1.0; python_version <= '3.12'"

(>2.1.0 is needed because of IGRF-14)

Thanks

[smithara]

I think it's better to specify simply apexpy>=2.1.0 for the optional dependency. Restricting it to only be installed when on a specific Python version makes it hard to test on the other Python versions without changing the pyproject.toml every time.

I've just tested installing apexpy on github actions and seems to be okay for version 3.9-3.13 inclusive, at least on Linux. Needs some extra steps for Windows/Mac (I think you've found the same). I'd suggest only running the minimal install across all platforms/versions, and run the "full" install (including apexpy etc), only on Linux. I'll put a PR together for this.

[ancklo]

Perfect! Thank you so much for the update #15