Use OS-specific cache directories instead of home directory #31295
Conversation
|
I think this would warrant a new entry, see: https://github.com/scikit-learn/scikit-learn/blob/main/doc/whats_new/upcoming_changes/README.md for instructions. There's quite a few other places that need to be updated, I would suggest searching for "scikit_learn_data" in the codebase. |
|
I believe I have covered all the files now |
|
Thanks for the PR @norgera. This is a breaking change so I think it needs a deprecation cycle. |
|
Indeed. Here is a link on how we want to go through when it comes to deprecation: https://scikit-learn.org/dev/developers/contributing.html#deprecation |
|
Thank you all for the information. I've added deprecation to use the original home directory if its detected or manually set, providing a warning + instructions to move files to new cache folder. Otherwise uses new cache folder. Let me know if this approach is appropriate. |
Resolves #31267
The get_data_home function now uses standard OS cache directories:
Previously, data was stored in ~/scikit_learn_data by default.
This change follows OS conventions for cache storage and improves
maintainability.
Implemented deprecation protocol and added tests in test_base.py