Jérôme Dockès

indeed the best way to get nibabel images into NiiVue is probably to serialize them as a Nifti file then pass that to NiiVue. writing them to a temporary location...

@hanayik thanks! here is the script that I had tried yesterday: ```python import string import base64 from nilearn import datasets from nilearn.plotting import html_document template = string.Template( """ NiiVue var...

note the way to plot an anat image with the current API is to use view_img. unlike the plot_ functions, there are no view_stat_map, view_anat, view_roi, etc but only view_img...

> Ok, so what do you think is the first method we should add / update in order to integrate niivue to nilearn? IMO building on the first example you...

I'm not against adding view_anat & friends, but as they would just be convenience functions to provide better defaults for specific image types, the main focus for niivue integration is...

https://github.com/jupyterlite/jupyterlite/issues/1023

after discussing a bit it's probably not worth the effort; installing from pypi is good enough for scikit-learn so it's probably good enough for skrub as well :)

note: this only happens if the the y Series doesn't have a `name`

@GaelVaroquaux is [this](https://output.circle-artifacts.com/output/job/e0bf48ba-4054-48a9-b929-b4bbb85eeced/artifacts/0/doc/auto_examples/00_getting_started.html#sphx-glr-auto-examples-00-getting-started-py) what you have in mind? I'm not sure which I like better, from the user's point of view it's not obvious why the filter they chose would...

In the future we might add more tabs for which filtering columns doesn't make sense; that would be an argument for the "gray-out rather than apply" option.