Algorithm animation, the use of dynamic graphics to illustrate program behavior and enhance understanding, has been successfully utilized as a pedagogical aid. Typically, algorithm animation involves post-mortem program augmentation: graphics are designed and developed for already working programs. The techniques used in algorithm animation systems could be adapted for use during program debugging. The objectives of this research are to use the abstract mental models developed by program designers as a basis for program visualizations and to develop a framework that allows programmers to quickly and easily (without graphics coding) create an animated visualization of a program prior to testing and debugging. Such a framework then could be used to examine program behavior and identify errors in a uniquely different manner than traditional textual debuggers. The intended deliverables include (1) a description and implementation of a methodology for mapping program data and actions to dynamic graphics, (2) a prototype animation-based visual debugger, and (3) an analysis of the effectiveness of such a system.
