Continuous software engineering and its support by usage and decision knowledge: : An interview study with practitioners

Continuous software engineering (CSE) emerged as a process that is increasingly applied by practitioners. However, different perceptions impede its adoption in industry. Furthermore, opportunities through utilizing usage and decision knowledge remain unexploited. We conducted a semi‐structured interview study with 24 practitioners from 17 companies to study how practitioners apply CSE during software evolution and how usage and decision knowledge can support CSE. Regarding the application of CSE, we identified five perspectives on CSE with tool‐ and methodology‐driven definitions most prevalent. Automated tests, involved users, and shared rulesets are perceived as most relevant for CSE. Practitioners report more positive than negative experiences; however, more than half of their responses were neutral. Practitioners' future plans for CSE focus on enhancement, expansion, and on‐demand adaption. Regarding the integration of usage and decision knowledge into CSE, practitioners perceive accountability and traceability as major benefits, while raising concerns about its feasibility and user groups. As short‐term extensions, practitioners expect improvements regarding automation and role aspects, while long‐term additions to integration and experimentation capabilities are demanded. We conclude that CSE remains partially difficult to capture for practitioners, while their attitude toward integrating usage and decision knowledge into CSE is positive.

Utilizing a semi‐structured interview study with 24 practitioners from 17 companies, we identify five CSE definitions. Practitioners perceive quality as most relevant. They provide 19 positive, 56 neutral, and 17 negative experiences and three strategies for future plans. We structure CSE in a model. Practitioners welcome the integration of usage and decision knowledge into CSE. They report on benefits, obstacles, additions, and extensions for both knowledge types, from which we improve a vision of knowledge integration into CSE.