Oh yeah, for sure it’s more robust.
But at the same time it still kinda hard to see the best approach for some cases that it’s not obvious to do the rollback.
For example, let’s say one of your operation is to replace the current value of a row(s) inside the database. After you replaced it, you lost the old information, so I’m not sure what would be the best approach here to restore the old data.
Maybe you can first get the old data and keep it in memory so if a rollback is needed you can replace it back with the old data. That would probably work, but maybe there is a better way.
