Automated design flaw correction in object-oriented systems

Abstract

Software inevitably changes. As a consequence, we observe the phenomenon referred to as "Software Entropy" or "Software Decay": the software design continually degrades making maintenance and functional extensions overly costly if not impossible. There exist a number of approaches to identify design flaws (problem detection) and to remedy them (refactoring). There is, however, a conceptual gap between these two stages: There is no appropriate support for the automated mapping of design flaws to possible solutions. In this paper we propose an integrated, quality-driven and tool-supported methodology to support object-oriented software evolution. Our approach is based on the novel concept of "correction strategies". Correction strategies serve as reference descriptions that enable a humanassisted tool to plan and perform all necessary steps for the safe removal of detected design flaws, with special concern towards the targeted quality goals of the restructuring process. We briefly sketch our tool chain and illustrate our approach with the help of a medium-sized real-world case-study

Measurement and quality in object-oriented design

“…In the years following the defense of the thesis, the concepts introduced here inspired the research of other Ph.D. students: Raţiu and Gîrba [17] applied the detection strategy concept for finding design problems that are revealed by the history of the system; while Trifu [19] defines correction strategies as a mean to address by refactorings the design problems which are detected using the suite of detection strategies defined in this dissertation.…”

Archimetrix: Improved Software Architecture Recovery in the Presence of Design Deficiencies

“…Trifu et al define quality goals and order detected flaws depending on their impact on these goals [27]. The calculated impact is however based on beforehand estimations of the influence a flaw has on a given quality goal.…”

Reengineering component-based software systems with Archimetrix

“…In [56], Trifu et al detect and remove design flaws with respect to a user-selected quality goal, e.g., performance. They detect those flaws by using graph matching in combination with basic metrics.…”