Feature Request
Currently, datasets are downloaded into the python environment of the current working directory. This can lead to unexpected behaviour and redundancy, especially across multiple virtual environments.
Proposal
Change the default dataset storage path to a more standard, centralized cache directory: ~/.cache/tgb.
This mirrors best practices for most projects which use ~/.cache for storing large files and avoiding duplication.
Additionally
Optionally override with an env var alike TGB_DATA_DIR for custom paths in case users are on a file-system where particular mount points are quick for disc reads (e.g. cluster). Fall back to default ~/.cache/tgb otherwise.
Feature Request
Currently, datasets are downloaded into the python environment of the current working directory. This can lead to unexpected behaviour and redundancy, especially across multiple virtual environments.
Proposal
Change the default dataset storage path to a more standard, centralized cache directory:
~/.cache/tgb.This mirrors best practices for most projects which use
~/.cachefor storing large files and avoiding duplication.Additionally
Optionally override with an env var alike
TGB_DATA_DIRfor custom paths in case users are on a file-system where particular mount points are quick for disc reads (e.g. cluster). Fall back to default~/.cache/tgbotherwise.