A Methodology for Increasing the Dependability of Open Source Software Components

Abstract

With the increasing use of open source software components in real world systems, there has arisen a need for increasing the dependability of these components. The dependability of these components can be increased like any other software component, by adding error detection and error recovery mechanisms to the code to deal with unforeseen errors and faults occurring in the system that will be running the software. The major issue is not in the implementation of these error handling mechanisms, but instead in the speed that they can be applied and where to apply them. The reason for considering an open source solution to a problem is most likely to save time and money using a solution that may already do what needs to be done. The goal is to incorporate these open source components into a system after verifying and if need be, increasing the dependability of the component. But, with open source components, they are already in a maintenance stage of development, where the code is either refined or new functionality is possibly added. Thus, standard techniques that rely on use during development will take too long to implement. This thesis proposes a novel way of determining the highest priority modules in the system and the most important locations for placing error handling mechanisms in the functions within those modules.