Every time something is updated or deleted, the entire previous record is inserted into its corresponding _A table with who did it and when (and optionally for what transaction number)
so delete from foo results in an insert into foo_A before the delete occurs.
so update to bar results in an insert into bar_A before the update occurs
and rolling back a delete results in picking the old values out of the _A table
and rolling back an update results in picking the old values out of the _A table and inserting the current values into the _A table before the updating the main table.
I guess there's plenty of technical solutions to track changes, like scd or snapshots or audit tables - what I find interesting about git approach it's that it's much more user-friendly vs. SQL archeology to understand what changed?
