Formalisms in Software Engineering: Myths Versus Empirical Facts

Abstract

Abstract. The importance of software grows in all sectors of industry and all aspects of life. Given this high dependability on software, the status of software engineering is less than satisfactory. Accidents, recall actions, and late projects still make the news every day. Many of the software engineering research results do not make it into practice, and thereby the gap between research and practice widens constantly. The reasons for not making it into practice range from isufficient commitment for professionalization of software development on the industrial side, to insufficient consideration for practical scale-up issues on the research side, and a tremendous lack of empirical evidence regarding the benefits and limitations of new software engineering methods and tools on both sides. The major focus of this paper is to motivate the creation of credible evidence which in turn will allow for less risky introduction of new software engineering approaches into practice. In order to overcome this progress hindering lack of evidence, both research and practice have to change their paradigms. Research needs to complement each promising new software engineering approach with credible empirical evidence from in vitro controlled experiments and case studies; industry needs to baseline its current state of the practice quantitatively, and needs to conduct in vitro studies of new approaches in order to identify their benefits and limitations in certain industrial contexts.

An industrial case study of classifier ensembles for locating software defects

“…Recent research proposed by Rombach et al (2008) points out that research and industry need to cooperate on current practices of software engineering in order to gain insight in how software development can be managed better. One of the reasons for current problems such as schedule and budget overrun, and system failures especially in safety-critical embedded applications is summarized as ''non-compliance with best practice principles of process design'' (Rombach et al 2008).…”

“…One of the reasons for current problems such as schedule and budget overrun, and system failures especially in safety-critical embedded applications is summarized as ''non-compliance with best practice principles of process design'' (Rombach et al 2008). These typical challenges exist in practical software engineering due to lack of empirical evidence regarding the benefits and limitations of existing approaches in this context.…”

“…Results obtained from effective methods or tools, i.e. early detection mechanisms, would definitely guide managers to baseline their state of practice in a quantitative manner rather than relying on mythical facts (Rombach et al 2008).…”

“…Based on the fact that software engineering research should also complement new approaches with empirical facts from industrial case studies (Rombach et al 2008), we define our research question Software Qual J (2011) 19:515-536 519 that would address the challenges in embedded software development. From an industrial perspective, software managers aim to decrease their testing efforts while decreasing defect rates, thereby producing high quality embedded software.…”

Test Generation Using Symbolic Animation of Models

“…• For the purpose of assuring the quality of classical software systems, low-level source code testing -though still necessary-is not the most important method of quality assurance any longer: Software engineers nowadays are (or, at least, should be) guided by the principle of 'correctness by construction' (Shukla et al 2010) which goes hand-in-hand with valuable and effective development techniques such as source code inspections (Rombach et al 2008). Those techniques, especially source code inspection, are usually not ( On these premises, we can now also make plausible why the main intent of this chapter is directed towards model-based, passive testing of non-software components, though software components will be mentioned as well.…”