Can import in conda environment but not associated jupyter notebook kernel

I'm writing a package, and the most recent update is reflected in my conda environment but not the associated jupyter kernel. An old version of the package is appearing there.

First, I've verified that the sys.executable and sys.path are the same between the command line interpreter and Jupyter Notebook. I've tried uninstalling the package by deleting the files created during installation, and I've tried to delete .ipynb_checkpoint files as well as __pycache__ files.

At this point, I've installed the most recent package in every single conda environment as well as root, and it works in all environments via the interpreters, but does not appear to be updated in any of the jupyter notebook kernels.

Any ideas?

Edit: the change in the most recent version of the package was simply to add the import of a subpackage to the __init__.py file. When I import the top level package and dir(package), I can see that the subpackage is missing in jupyter notebook but not in my command line interpreter.

Start by comparing sys.executable and sys.path between the command-line interpreter and the Jupyter Notebook kernel, then inspect the package's __init__.py and its installed files. Check whether the notebook still imports an old package version; done means the kernel exposes the updated top-level package and its newly imported subpackage.

Written by the indexing model from the issue text.