Reconstructing higher level change information from versioning data

Abstract

Change is an essential element of the software development cycle, certainly since systems which do not change, perish. Despite the importance of changes, most evolution studies evaluate the impact of changes by comparing two or more snapshots of a system over time instead of reconstructing and evaluating the actual change operations. Therefore we propose a number of techniques which are capable of reconstructing change-operations from the low-level data stored in versioning systems in order to learn more about the current system as well as the change process. More specifically, we show how these techniques can be used to study the use of move-operations