Identifying refactoring sequences for improving software maintainability

Abstract

Refactoring is a well-known technique that preserves software behaviors and improves its bad structures or bad smells. In most cases, more than one bad smell is found in a program. Consequently, developers frequently apply refactorings more than once. Applying appropriate refactoring sequences, an ordered list of refactorings, developers can remove bad smells as well as reduce improvement time and produce highly maintainable software. According to our 2011 survey, developers consider four main criteria to select an optimal refactoring sequence: 1) the number of removed bad smells, 2) maintainability, 3) the size of refactoring sequence and 4) the number of modified program elements. A refactoring sequence that satisfies these four criteria produces code without bad smells, with higher maintainability, using the least improvement effort and time, and providing more traceability. Some existing works suggest a list of refactorings without ordering, and others suggest refactoring sequences. However, these works do not consider the four criteria discussed earlier. Therefore, our research proposes an approach to identify an optimal refactoring sequence that meets these criteria. In addition, it is expected that the findings will reduce maintenance time and cost, increase maintainability and enhance software quality.