Abstract:

Kernel defects in operating systems make up around 40% of the overall defects based on data collected over more than 14 years for three successive operating system releases (R8, R9 and R10). The study shows that the time it takes each OS release to stabilize increases despite the reusability of the code and the experience gained by programmers in developing previous releases. Many of the defect categories observed in one release show up in subsequent releases, although one expects that the old fixes, batches, and corrections would contribute to a cleaner newer release. The percentage of new defect types observed in subsequent releases is small compared to previously encountered defect types. The data presented in this paper shows the time dependent defect arrival rate. This data helps a better planning and utilization of resources in order to improve the overall productivity.