Karlsruhe, February 2011 - The German Testing Board (GTB) put on an expert discussion on 23rd February 2011 during the "Software Engineering 2011" conference in Karlsruhe.
The discussion was moderated by Prof. Dr. Ernst Denert, Vice President of the Gesellschaft für Informatik. From the podium Klaus Beetz (Siemens AG), Prof. Dr. Mario Winter (Cologne University of Applied Sciences), Dr. Valentin Dallmeier (Saarland University), Prof. Andreas Spillner (Bremen University of Applied Sciences) and Monika Bögge (Telefónica o2 Germany) discussed why software has so many defects, and what can be done to avoid these if possible. An audience of more than 40 took an active part in the discussion.
„The German Testing Board does not operate in isolation. We try to cooperate closely with industry, consultancies and software application companies as well as with universities. The purpose of an expert discussion is to bring all parties together aound one table and to enable an exchange of views about new developments. We as the GTB also want to take the opportunity at such events to broaden our knowledge – for example regarding improvements to the syllabi“, says GTB Chairman Tilo Linz following the event.
During the expert discussion practitioners and researchers explored the subject from different angles. The central questions were:
- How can defects be found at an early stage?
- What is used by industry and what does industry demand?
- Which methods and techniques are state-of-the-art?
Google recently awarded a prize worth 750,000 USD to the chair in software engineering of the Saarland University for its research work.
Dr. Valentin Dallmeier, post-doctorate fellow at this faculty, represented the department chair Prof. Dr. Andreas Zeller and took part in the expert discussion. He explained the basic approach of the Saarland University in connection with the Google research assignment: „The manual work that testers have to perform should be increasingly automated. The idea is to combine different scientific research subjects so that better, shorter and more efficient tests can be performed. The test generation techniques, which have existed in the research for some time, should be transformed into practicable real-world solutions, so that industry can benefit from our research.“ In another context, Prof. Zeller also describes his research work as follows: „Traditional software tests function much like searching for a needle in a haystack. You will find something in different areas, but you can never be sure whether you have actually found all defects.”
The two representatives from industry, Klaus Beetz and Monika Bögge, were then asked whether in their respective organizations research results are actually already applied in practice, and which methods/techniques/tools their organizations would wish to have in future. „Testers are for me the central figures in each software project because they have to accomplish two things: On one hand they have to understand the project from the user side, and on the other hand they also have to grasp the internal structure of the project, e.g., for testing non-functional attributes such as performance or scalability. I support the Certified Tester training scheme because it teaches testers good standard skills”, said Klaus Beetz from Siemens AG.
Prof. Andreas Spillner pointed out how significant communication in test processes is: „Particularly in software development, all stakeholders must talk about the processes in order to achieve clarity. Only in the wake of this can the quality of the product be increased through the reduction of defects.“
Prof. Mario Winter said that testers must be capable of understanding not only the purely business-related but also the technical aspects of software development. “Then they can contribute to improving the quality of software during its development.“
„The devil takes the hindmost!“ With this witty remark, moderator Prof. Ernst Denert drew attention to an inherent problem in software testing. „It still occurs too often that testing is the last step of a process. The effort for testing is then in most cases calculated too tightly since the customer’s deadline looms. It is an important responsibility of the project leader to take the necessary time for testing.“

