The author makes a case against using “valid_from”/“valid_to” columns on your table. But in my experience, that is the only viable solution. Fx joining an employee-table and a department-table needs to take into account when an employee was in which department. How would do that with some git-like version control for data? That might be useful for some machine learning training dataset, but in an operational system? No thanks
Hi! Author here. The git model lets you track multiple tables in the same repo. You can query (and join) both tables starting with the same commit_id and is actually a nice solution to the problem you suggest :)