A Quality Driven Approach to Improve Legacy Code Projects

Abstract

In practice, re-use and further development of existing software products is often harder to achieve than initially planned. Some specific reasons are unavailability of initial developers, lack of documentation, and focus on supporting one project or mission. The core problem can be summed up with: “It has not been built to react on changing requirements.” In such a case, developers often neglected internal software quality. Symptoms include tightly coupled software modules or absence of a regression test suite consisting of fast unit tests. In the end, it is really hard to add new functionalities without introducing unforeseen side effects. But how can you proceed with such a legacy code project? Beginning development from scratch is often not feasible. In this talk, we present a novel quality driven approach to improve internal software quality while keeping the project active for further development. We outline basic steps such as initial analysis, creation and prioritization of the refactoring roadmap, or setting up the regression test environment. Particularly, we want to show how to practically approach large code refactoring iteratively. We illustrate all steps and report our lessons learned at the example of a distributed, workflow-driven integration framework and a library for orbit analysis.