Automated Assistance for Search-Based Refactoring Using Unfolding of Graph Transformation Systems

摘要

Refactoring has emerged as a successful technique to enhance the internal structure of software by a series of small, behaviour-preserving transformations [4]. However, due to complex dependencies and conflicts between the individual refactorings, it is difficult to choose the best sequence of refactoring steps in order to effect a specific improvement. In the case of large systems the situation becomes acute because existing tools offer only limited support for their automated application [8]. Therefore, search-based approaches have been suggested in order to provide automation in discovering appropriate refactoring sequences [6,11]. The idea is to see the design process as a combinatorial optimization problem, attempting to derive the best solution (with respect to a quality measure called objective function) from a given initial design [9].