Closed pbellec closed 2 years ago
About user storage, i finally found that binderhub (or rather jupyterhub) allocate 10GB per user by default. https://zero-to-jupyterhub.readthedocs.io/en/latest/jupyterhub/customizing/user-storage.html
For the dataset storage we agreed on few GB.
For what folder is acessible or not, we could use cookiecutter data layout proposal https://github.com/drivendata/cookiecutter-data-science/tree/master/%7B%7B%20cookiecutter.repo_name%20%7D%7D/data
REPO_ROOT/data/raw
would be read-only, where all neurolibre datasets are stored.
the rest can be user by the user.
https://github.com/neurolibre/neurolibre-reviews/issues/7#issuecomment-1006921585
and encourage pulling from stable sources (inc DOIs).