Understanding-in-the-large

Developing and maintaining large industrial software products implies programming-in-the-large activities. Related concepts are usually represented in terms of low level features such as file system hierarchies, preprocessor files, makefiles, shell scripts, sccs archives, etc. Understanding the information embedded in such artifacts is an important but difficult task, especially with neither conceptual framework, nor tool assistance. To emphasize the importance of this issue, this paper makes the distinction between understanding-in-the-large and understanding-in-the-small. Using a conceptual classification, understanding-in-the-large problems are described in a structured way. Difficulties in building reverse-engineering-in-the-large tools are then analyzed and illustrated taking preprocessor files as a case study. The Champollion approach to these problems is briefly presented.