Automated Regression Hunting

Code regressions happen, it is almost a truism, and the more complex the system, the more often they will occur. Detecting which code addition or patch created the regression, while sometimes time-consuming, is at heart an iterative process which lends itself to automation. The core of this process is the same regardless of the kind of software being written, whether it is application programming or operating system programming.

The PyRet project has automated this process in an object-oriented fashion using Python. It includes an implementation to test the Linux Kernel using the Linux Test Project.

This paper will describe the process of regression hunting which lies at the heart of the PyRet project. It will provide: a basic overview of the PyReT architecture, the search models that PyReT currently utilizes, and methods for extending the PyReT tool to other software projects.


Download PDF.