Dependency updates

[jrdalenberg]

I was doing a power analysis for myself and noticed that the code was not working with the latest packages.

These updates work for me using Python 3.9.18.

I updated:

• appending pandas dataframes with new dataframes
• Nibabel data extraction
• Mask calculation by using math_img from Nilearn.
• Matplotlib normed argument.
• The example notebook

[Remi-Gau]

Hey @jrdalenberg

Thanks for the PR.

Need to double check but:

• plotting seems only used in the notebook so not sure it is worth adding a dependency to the package
• similarly I wonder if we need to have all of nilearn as a dependency for this much

Making sure that the demo notebook can run at any time is VERY good point though so it should probably be part of the CI.

[Remi-Gau]

Also it seems that the changes are breaking some tests on Windows only.

[Remi-Gau]

Also it seems that the changes are breaking some tests on Windows only.

my bad the tests are failing on main too, so this is not related to the PR. Sorry.

[jrdalenberg]

Hey @jrdalenberg

Thanks for the PR.

Need to double check but:

* plotting seems only used in the notebook so not sure it is worth adding a dependency to the package

* similarly I wonder if we need to have all of nilearn as a dependency for this much

Making sure that the demo notebook can run at any time is VERY good point though so it should probably be part of the CI.

Thanks for the responses.

@ Plotting: matplotlib was already in there. The line was only updated with a comma. The plotting features in core code used an outdated argument.

@ Masking using nilearn: I understand your point. If I find some time I could implement just the masking logic.

Should I wait on a fix for the windows tests?

[Remi-Gau]

@ Plotting: matplotlib was already in there. The line was only updated with a comma. The plotting features in core code used an outdated argument.

And that's why I should not review things too early in the morning!!! 🙈

@ Masking using nilearn: I understand your point. If I find some time I could implement just the masking logic.

If you have time that'd be great

Should I wait on a fix for the windows tests?
nope: keep going without a fix for those

[Remi-Gau]

FYI I have just updated the code formatting tooling.

To get rid of the annoying CI failure you can:

• merge the upstream/master branch into your branch https://github.com/jrdalenberg/neuropower-core/tree/dependency-updates
• install pre-commit: pip install pre-commit
• check all files for formatting and linting: pre-commit run --all-files
• commit the changes