pooch icon indicating copy to clipboard operation
pooch copied to clipboard

Choose a more obvious system cache dir when using pooch within Google CoLab

Open patcon opened this issue 2 weeks ago • 1 comments

Description of the desired feature: When using pooch in Google CoLab, set the system cache to /content/sample_data rather than the default hidden /root/.cache (which is a bit strange to find if you aren't familiar with the steps to show hidden files in CoLab)

See: https://stackoverflow.com/a/53586419

import sys
IN_COLAB = 'google.colab' in sys.modules

Are you willing to help implement and maintain this feature?

I'm new to the ecosystem, but this seems simple enough to own responsibility for :)

The SO answer hasn't changed since 2019, so it seems like the solution is quite stable.

Thanks for the great tool! I'm now down a rabbit hole of thinking about doi's, and generally considering my independent research project in a more formal way!

patcon avatar Dec 11 '25 21:12 patcon

I'm seeing that this would very intentionally be storing files in the user_data_dir, which is maybe a non-starter.

I suppose the main case for doing something very different in colab is because it's essentially an ephemeral filesystem, and so there's very little benefit in putting the data anywhere non-obvious, since it's only being cached very temporarily?

https://github.com/tox-dev/platformdirs?tab=readme-ov-file#platformdirs-to-the-rescue

patcon avatar Dec 11 '25 21:12 patcon