A Pilot Study of Interactive Versus Noninteractive Debugging

摘要

The experiment described in this document was conducted to obtain data for a larger study to determine the economics of different types of computer systems (time-sharing, batch-processing, personally operated or stand-alone, and multiple-console, open-shop). The data were gathered to permit an analysis of the degree to which the use of different types of computer systems changes the cost and effectiveness of program debugging. This exploratory experiment used nine SDC trainees who were in the latter phase of their Programmer Training Course. Each subject coded and debugged one problem on a time-sharing system (interactive), and another problem on a simulated multiple-console, open-shop system (noninteractive). Problems and subjects were counterbalanced in a Latin-Square experimental design. Statistically significant differences were found between the two systems, with better performance in the interactive mode. However, it is not felt that these results should be extended beyond the experimental situation. Large differences between individual programmers and difficulties in defining and measuring job characteristics have led to the recommendation that further study be directed towards more objective determination of the characteristics of the user (programmer), the jobs, and the system used. Only then can an attempt be made to determine how these three interact in trying to do a specified job for some type of user. (Author)