Selecting which tests to run can be difficult. Do we have too many tests? Do we have too few tests? We will look into two different ways to address this problem
*Test selection based on historical analysis*
Regression testing is done to see if introducing modifications to software affects existing functionality. The simplest regression testing strategy is to rerun all test cases. It is an easy strategy to implement, but often it is also unnecessarily expensive, especially if changes affect only a small part of the whole system under test. Therefore, applying regression test selection techniques can be of great advantage. Edward Ekelund will describe the results of his Master Thesis on test selection, which was done at Axis.
*High volume automated testing (HiVAT)*
Faced with new features, the amount of possible tests is usually enormous and attempting complete test coverage is often impractical if not impossible. Exploring user scenarios is one useful way of approaching this problem. This can be enhanced significantly by generating a large amount of combinations, run them automatically and flag for problems so a human can take a closer look. Baldvin Gislason Bern will describe how we have used this approach at Axis.
Edward Ekelund works with development tools at Axis Communications
Baldvin Gislason Bern works with testing at Axis Communications