Classification of Scattered Code in Object-Oriented Applications

Aspect-oriented programming strives to encapsulate crosscutting concerns into modular units called aspects. Aspect Mining techniques aim to search for aspect candidates in existing software systems. These techniques associate all occurrences of scattered code in existing applications with the absence of aspects. In this paper, we demonstrate that in a poorly designed object-oriented system, scattered code can also originate from the absence of objects. On the one hand it will be erroneous to associate the occurrences of scattered code originating from the absence of objects with the absence of aspects. On the other hand such scattered code should be separately classified as it identifies refactoring opportunities. In this paper, we first present patterns that are similar to the patterns searched by aspect mining techniques but these patterns originate from the absence of proper object-oriented design. Based on these patterns, we construct a tool that distinguishes between scattered codes arising due to the absence of different encapsulating units: objects and aspects. Hence, the tool differentiates between aspect-oriented and object-oriented refactoring opportunities. We validate our findings on different software systems. Keywords-component; Aspect-oriented Programming; Aspect Mining; Software Design Defects; Code Smells; Refactoring