We report the results of a survey about the software faults encountered during the testing phases in evolving a large real-time system. The survey was done in two parts: the rst part surveyed all the faults that were reported and characterized them in terms of general categories; the second part resurveyed in depth the faults found in the design and coding phases. For the rst part, we describe describe the questionaire, report the general faults found, and characterize the requirements, design and coding faults by the testing phases in which they were found and by the time they were found during the testing interval. For the second part, we describe the questionaire used to survey the design and coding faults, report the faults that occurred, how di cult they were to nd and x, what their underlying causes were that is, what their corresponding errors were, and what means might have prevented them from occurring. We then characterize the results in terms of interface and implementatio...
Dewayne E. Perry, Carol S. Stieg