An Empirical Approach to Software Archaeology

The term "software archaeology" provides a useful metaphor of the tasks that a software developer has to face when performing maintenance on large software projects. The source code of a program at any point in time is the result of many different changes performed in the past, usually by several people, which can be tracked when a version control system is used. We have designed a methodology for analyzing with detail the age of the source code in such cases, and have applied it to several large software projects. As a part of the methodology, we define a set of indexes which can help to characterize the history of a software system, and discuss how those could be used to estimate its past and future maintenance. We also show how our approach to software archaeology is simple both conceptually and computationally, but still very powerful at uncovering useful information.