| Popis: |
Background: Logs are the footprints that software systems produce during runtime, which can be used to understand the dynamic behavior of these software systems. To generate logs, logging practice is accepted by developers to place logging statements in the source code of software systems. Compared to the great number of studies on log analysis, the research on logging practice is relatively scarce, which raises a very critical question, i.e. as the original intention, can current logging practice support capturing the behavior of software systems effectively? Aims: To answer this question, we first need to understand how logging practices are implemented these software projects. Method: In this paper, we carried out an empirical study to explore the logging practice in open source software projects so as to establish a basic understanding on how logging practice is applied in real world software projects. The density, log level (what to log?) and context (where to log?) are measured for our study. Results: Based on the evidence we collected in 28 top open source projects, we find the logging practice is adopted highly inconsistently among different developers both across projects and even within one project in terms of the density and log levels of logging statements. However, the choice of what context the logging statements to place is consistent to a fair degree. Conclusion: Both the inconsistency in density and log level and the convergence of context have forced us to question whether it is a reliable means to understand the runtime behavior of software systems via analyzing the logs produced by the current logging practice. |
