An interview study about the use of logs in embedded software engineering

Context Execution logs capture the run-time behavior of software systems. To assist developers in their maintenance tasks, many studies have proposed tools to analyze execution information from logs. However, it is as yet unknown how industry developers use logs in embedded software engineering. Objective In this study, we aim to understand how developers use logs in an embedded software engineering context. Specifically, we would like to gain insights into the type of logs developers analyze, the purposes for which developers analyze logs, the information developers need from logs and their expectation on tool support. Method In order to achieve the aim, we conducted these interview studies. First, we interviewed 25 software developers from ASML, which is a leading company in developing lithography machines. This exploratory case study provides the preliminary findings. Next, we validated and refined our findings by conducting a replication study. We involved 14 interviewees from four companies who have different software engineering roles in their daily work. Results As the result of our first study, we compile a preliminary taxonomy which consists of four types of logs used by developers in practice, 18 purposes of using logs, 13 types of information developers search in logs, 13 challenges faced by developers in log analysis and three suggestions for tool support provided by developers. This taxonomy is refined in the replication study with three additional purposes, one additional information need, four additional challenges and three additional suggestions of tool support. In addition, with these two studies, we observed that text-based editors and self-made scripts are commonly used when it comes to tooling in log analysis practice. As indicated by the interviewees, the development of automatic analysis tools is hindered by the quality of the logs, which further suggests several challenges in log instrumentation and management. Conclusions Based on our study, we provide suggestions for practitioners on logging practices. We provide implications for tool builders on how to further improve tools based on existing techniques. Finally, we suggest some research directions and studies for researchers to further study software logging.