The dataset primary key columns are immutable now to avoid users messing up with the primary keys by performing illegal operations on those columns such as changing their types and deleting the columns.
To discuss: we should have versioning for the dataset and design a way to keep track of a stream of changes and operations from one version to the next.
The dataset primary key columns are immutable now to avoid users messing up with the primary keys by performing illegal operations on those columns such as changing their types and deleting the columns.
To discuss: we should have versioning for the dataset and design a way to keep track of a stream of changes and operations from one version to the next.