Lessons from a Domain-Based Reengineering Effort

We present in this paper the lessons and insights learned from of a domain-centered reengineering effort. Using a method we developed in a previous work, we set about to understand and transition a complete legacy system from COBOL to an executable domain model. Our work suggests that a domain-based approach is very promising but a number of issues remain to be better understood. Among these are questions about domain completeness, scoping, interleaving and evolution; concept matching at the granularity of both the programs' architecture and the details of the source-code; thoroughness and representation of the legacy programs coverage, as well as the problems inherent to the transition of a multi-programs system. We discuss these issues in details using examples. Implications on future work in the area are suggested.