Towards effect preservation of updates with loops

Todays technologies are capable to assist database developers in designing complex actions (e.g. stored procedures, active database rules). However, automatic generation and extention of given actions holds a danger - The original intention of actions might get canceled or even reversed, e.g. by some appended repairing action enforcing an integrity constraint. This problem might cause non-desired situations where an action commits although one of the intended database modifications was not applied. In this paper, we deal with the characterization and preservation of effects (intentions) of database actions. As an extention of our current approach for effect preservation, we present a method for handling updates including non-nested loops. A transformation process is proposed that modifies a given action S to S′, such that S′ does preserve the effects. In order to reduce additional run-time overhead of S′, most of the computation is shifted to compile-time.