High interaction data visualization using Seesoft to visualize program change history (abstract)

Abstract

A problem in developing large software systems is understanding the source code. This problem is difficult because of the volume of code. The listing for a moderately sized system with 100,000 lines, printed 50 lines per page, would run 2,000 pa~s. This video shows a new software tool, Seesoft , that applies scientific visualization techniques to visualizing code. The visualization approach is to represent files in a directoty in columns and the source code lines as rows of colored pixels. The indentation and length of each row of pixe18 corresponds to the actual code. The color of each row of pixels is determined by a statistic such as the age, programmer, or type of line, that we obtain from the change management system. The visual impression is that of a miniature picture of the source code with the indentation showing the usual C controls structure and the color showing the spatial distribution of the statistic. A user may adjust the display using direct manipulation techniques to discover interesting patterns in the code. Software engineering concepts such as complexity and bug fix on fix density can be visualized. The main interest of this work to the human factors community is the use of graphical user interface for selecting and combining statistics from a database, the effective use of hundreds of colors to display a mass of data, and the reduction of the point-and-click direct manipulation metaphor to just pointing, e.g. something of interest will occur where ever the mouse points to on the display.