So far, the datasets consist of a json file with a list of absolute paths. It means that they can't be shared between different users, if the root of the data folder is located at a different absolute path.
One way would be to list the paths in dataset relative to the data folder root, and maybe also store this data_root just for future information. When someone loads a dataset, HAL only reads the relative paths, and adapt them to the local root
So far, the datasets consist of a json file with a list of absolute paths. It means that they can't be shared between different users, if the root of the data folder is located at a different absolute path.
One way would be to list the paths in dataset relative to the data folder root, and maybe also store this data_root just for future information. When someone loads a dataset, HAL only reads the relative paths, and adapt them to the local root