There are many graphical techniques for visualizing sofiware. Unfortunately, the current techniques do not scale to display large software systems and are largely unused. We present a method for visualizing statistics associated with code that is divided hierarchically into subsystems, directories, and files. Using our technique, we can display the relative sizes of the components in the system, which components are stable and which are changing, where the new functionality is being added, and identify error-prone code with many bug fixes. Using animation, we can display the historical evolution of the code.

This paper describes a graphical technique implemented in a software tool, SeeSys, for displaying large volumes of software. The SeeSys visualization technique can be applied to display code for any existing software system organized hierarchically into subsystems, directories, and files. The idea is to represent the whole system using a rectangle and recursively represent each subunit wit h interior rectangles. The area of each rectangle is proportional to a statistic associated with its subunit, for example, NCSL. SeeSys then uses color, zooming, a fill statistic, and interactive techniques that layer additional information onto the base display.