One more problem discovered while preparing the v1.5 release: The bulk downloader (snewpy.get_models()) broke when moving model files to separate repositories.
This PR fixes it and ensures it uses the same download code as model initialisers (iterating over all parameter_combinations, rather than using its own dictionary of hardcoded file names); to avoid future breakage.
As a side effect, the bulk downloader will now download to the same snewpy.model_path, rather then the current directory. This is good for consistency and may avoid duplicate downloads.
The parameter download_dir was never officially documented; I’m still marking it as deprecated though, to be safe.
One more problem discovered while preparing the v1.5 release: The bulk downloader (
snewpy.get_models()
) broke when moving model files to separate repositories.This PR fixes it and ensures it uses the same download code as model initialisers (iterating over all parameter_combinations, rather than using its own dictionary of hardcoded file names); to avoid future breakage.
As a side effect, the bulk downloader will now download to the same
snewpy.model_path
, rather then the current directory. This is good for consistency and may avoid duplicate downloads. The parameterdownload_dir
was never officially documented; I’m still marking it as deprecated though, to be safe.