Refactoring Berkeley DB Using Aspect-Oriented Programming

摘要

Aspect-Oriented programming (AOP) is a newly developed technology which provides a graceful solution to the code crosscutting problem in the OO (object-oriented) system. In an OO system, the codes belonging to one module always scatter to other modules, which makes the system customization difficult to carry out. That's an instinct of OOP due to its incomplete encapsulation. AOP uses a linguistic unit called "Aspect" to make the crosscutting codes localized in one single place to eliminate the ripple effect of changing these codes. Database management system (DBMS) is a complicated software system with many interactive modules in its architecture. It's widely used in many different areas which demands we can easily make a DBMS customizable to meet the needs of different application backgrounds. In this paper, we propose a technical implementation on how to convert an OOP system to AOP system. We first introduce 5 kinds of extracting rules and bring out an extracting methodology based on the location of the crosscutting code. Next we use Berkeley DB (BDB) as a case study to demonstrate how the rules can help us to make a DBMS tailorable so as easier to customize. Through our experiment we determine that AOP is a very efficacy way for separating crosscutting concerns. It can improve the modularity and maintainability of the original system. But there still existing some disadvantages and drawbacks about this technology. We also evaluate these challenges in this paper.