Visual testing of software

Abstract

Software development is prone to time-consuming and expensive errors. Finding and correcting errors in a program (debugging) is usually done by executing the program with different inputs and examining its intermediate and/or final results (testing). The tools that are currently available for debugging (debuggers) do not fully make use of potentially useful visualisation and interaction techniques.

This thesis presents a new interactive graphical software testing methodology called visual testing. A programmer can use a visual testing tool to examine and manipulate a running program and its data structures.

Systems with techniques applicable to visual testing in the related domains of debugging, software visualisation and algorithm animation are surveyed. Techniques that are potentially useful to visual testing are described, examined and evaluated, and a design for a visual testing tool based on these techniques is presented. The tool combines aspects of user-controlled algorithm simulation, high-level data visualisation and visual debugging, and allows easier testing, debugging and understanding of software.

A prototype visual testing tool is presented and evaluated here as a proof of concept for some of the aspects of visual testing. Finally, some suggestions for future research in visual testing are presented.

Download

Visual testing of software (gzipped PostScript 2.0)

Visual testing of software (PDF 1.3)