We should save checksums for all derives datasets whether they are saved or not. When saved derived data are updated, or when conversion scripts or dockerfiles are updated, the new checksums should be automatically compared with the old ones to make sure that the data haven't changed.
We should save checksums for all derives datasets whether they are saved or not. When saved derived data are updated, or when conversion scripts or dockerfiles are updated, the new checksums should be automatically compared with the old ones to make sure that the data haven't changed.