A formal model of re-execution in software process

Abstract

Redoing has been introduced as a fundamental mechanism to handle the dynamics and flexibility required in the software process. It is an operation that involves canceling a part of a process enaction that is polluted by erroneous and incomplete activities and doing that part again. In order to make redoing effective, it is essential to detect the cause of errors correctly. A functional model that makes it easy to do this by dependency analysis is presented. A formal semantics of redoing this metaoperations which can handle the computational history as data is given. Its effectiveness is shown using the ISPW6 example process. Some extensions to minimize the cost of reexecution and to create scripts incrementally are proposed.<>