Evaluating the Effort for Modularizing Multiple-Domain Frameworks Towards Framework Product Lines with Aspect-oriented Programming and Model-driven Development

摘要

Multiple-Domain Frameworks (MDFs) are frameworks that unconsciously involve variabilities from several domains and present two main problems: i) useless variabilities in the final releases and ii) architectural inflexibility. One alternative for solving this problem is to convert them into Framework Product Lines (FPL). FPL is a product line whose members are frameworks rather than complete applications. The most important characteristic of FPLs is the possibility of creating members (frameworks) holding just the desired variabilities. However, the process of converting an MDF into an FPL is very time-consuming and the choice for the most suitable technique may improve significantly the productivity. The main focus of this paper is an experiment that evaluates two techniques that are usually considered for dealing with features: model-driven development and aspect-oriented programming. Our experiment was conducted comparing the effort in converting an MDF called GRENJ into an FPL called GRENJ-FPL The results showed significant differences regarding the time spent and the occurrence of errors using both techniques.