We should have some sort of end-to-end testing to ensure a new dataset revision doesn't introduce breaking changes so we are notified if a dataset stops being updated or if the data within it radically changes (i.e. new columns, new unexpected files, way smaller files, different files with wrong names)
We should have some sort of end-to-end testing to ensure a new dataset revision doesn't introduce breaking changes so we are notified if a dataset stops being updated or if the data within it radically changes (i.e. new columns, new unexpected files, way smaller files, different files with wrong names)