Event-Driven Metrics

This paper describes the notion of event-driven metrics as an alternative to generating metrics from software project artefacts on-demand or periodically. The rationale for wanting such an alternative stems from on-demand metric updates being prone to human error and for potentially missing important project events and trends. Furthermore, periodic updates, even if they are automatic, can also miss data and, at the other extreme, may even become burdensome if their frequency is too high. We begin by using scenarios to show how the event-driven mechanism might work, and then highlight some of the implementation issues. Finally, we point to the OPHELIA project as a concrete implementation of this notion.