The current implementation of update_duplicates is such that it updates all columns if there is a conflict during the insert phase. This is not always desirable. A classic case is the created_at.
If an insert fails, and we want the update to happen instead we would typically want created_at to keep its existing value. An option to control this behaviour would be great.
The current implementation of
update_duplicatesis such that it updates all columns if there is a conflict during the insert phase. This is not always desirable. A classic case is thecreated_at.If an insert fails, and we want the update to happen instead we would typically want
created_atto keep its existing value. An option to control this behaviour would be great.