Past and Future of Software Testing and Analysis
1 like524 views
The panel discusses the past and future of software testing and analysis. In the last decade, contributions included search-based software testing and using machine learning in test automation. Future directions include focusing on practical and scalable solutions, combining technologies like machine learning and symbolic execution to solve specific problems, and conducting applied research driven by industrial contexts. Impact requires collaboration between researchers and industry to understand problems.
More Related Content
![[2017/2018] RESEARCH in software engineering](https://cdn.slidesharecdn.com/ss_thumbnails/ivano01seresearch-171120182253-thumbnail.jpg?width=560&fit=bounds)
![[01-B] Empirical software engineering](https://cdn.slidesharecdn.com/ss_thumbnails/01-bempiricalsoftwareengineering-170904122621-thumbnail.jpg?width=560&fit=bounds)
Similar to Past and Future of Software Testing and Analysis (20)
Past and Future of Software Testing and Analysis
- 1. Panel: Past and Future of Software Testing and Analysis Lionel Briand http://www.lbriand.info ISSTA 2021
- 2. Contributions in the last decade • Search-based software testing • E.g., scalable test data generation, e.g., at system level • Use of Machine learning in test automation • E.g., Regression test selection and prioritization • Use of test execution history and change data across builds • In continuous development and integration environments • Testing cyber-physical systems • Testing CPS models (falsification), simulation-driven testing • Trace analysis and monitoring, e.g., specification languages 2
- 3. Future directions • We have developed many interesting ideas and concepts • Many of them are not practical or scalable, in most contexts • For example: mutation testing, metamorphic testing, etc. • We need to focus on devising novel engineering solutions that are scalable and practical, at least in some well- defined contexts • E.g., mutation analysis in embedded systems in the space domain 3
- 4. Future directions • Multi-disciplinary research: It is rarely the case that one technology solves an actual problem • We have many performant technologies at our disposal, that have made big leaps forward in the last decade: • Machine learning • Natural Language Processing • Meta-heuristic search • Solvers (e.g., SMT) • Symbolic execution • Simulation • How to combine them in a specific context for targeted problems? This is hard work, e.g., Automotive DS. 4
- 5. Critical factors for the field • We need fundamental research, not driven by specific problems, and we have done that well as a community • However, our practical impact has been limited and, over time, this is being noticed by institutions, funding agencies, etc. • Other fields have had much more (visible) impact • Testing and analysis research is a practical field that should have substantial visible impact 5
- 6. Critical factors for the field • Impact requires a thorough understanding of problems • Present and foreseeable problems • Problem understanding requires collaboration between researchers and industry • Collaboration needs to be supported and encouraged by our community, academic institutions, and funding agencies 6
- 7. Applied research, driven by industrial contexts, focused on scalable and applicable solutions, is difficult and rewarding research work 7
- 8. Panel: Past and Future of Software Testing and Analysis Lionel Briand http://www.lbriand.info ISSTA 2021