The agent said 1 row. The database changed 4. Caught by Deedglass.
The task: “Delete order 42.” The agent is scripted; the database result is a real run on the /play store.
The run
ClaimedWhat the agent says
DELETE FROM orders WHERE id = 42
“Removed order 42, 1 row.”
1row
ProvenWhat the database committed
public.orders, 500 rows. Order 42 is outlined.
4rows
- 1 row deleted in orders
- 3 rows deleted in order_items
- It ran. The tool reported 1 row.
Doesn’t match
The agent said 1 row. The database changed 4 rows.
The hidden catch: A cascade. A foreign key set to ON DELETE CASCADE removes rows in other tables too.
This copy of the store deletes an order’s line items and refunds with it (ON DELETE CASCADE).
In our benchmark, the same kind of case (“A foreign key cascade deletes 12 rows in another table”): Deedglass caught after it ran.
Share it.
