Quality requirements in agile as a knowledge management problem: More than just-in-time

Just-in-time (JIT) approaches have been suggested for managing non-functional requirements in agile projects. However, many non-functional requirements cannot be raised and met on the spot. In this position paper, we argue that effective JIT engineering of quality requirements depends on a solid foundation of long-term knowledge about all relevant quality requirements. We present two examples from projects related to safety and security and show that not all aspects of these quality requirements can be invented and changed just in time. Further, managing, for example, operationalization of quality requirements just in time depends on sufficient understanding of (i) customer value and (ii) the system under construction that must be shared by the engineering team. If a Learning Software Organization (LSO) intends to increase agility and speed up system development, it needs a holistic concept for managing this knowledge. We propose that a knowledge-management framework can facilitate JIT-RE by structuring, representing, and allowing updates of long-term knowledge about quality requirements. Such a knowledge-management framework should allow to map user value to system requirements and have important properties to allow JIT RE and sustainable evolution.