Increasing Software Quality using the Provenance of Software Development Processes

Abstract

Today’s software development processes are complex. A lot of interaction occurs between developers, the tools they use, and even automatically between different tools. Examples of those interactions are entering a new requirement into the bug tracking system, committing new source code to the repository or automatic code style check during a check-in. To trace and understand the full process is hard. To get insight into these processes and to increase the quality of the resulting software release, we record information about the process during run-time. This information is called the Provenance of the process. With Provenance we can analyze and audit the software development process with the goals error detection, quality assurance, process validation, monitoring, statistical analysis, process optimization, or developer rating. For example, we can answer question such as “Which requirement causes most bugs in release version X?” or “How many commits did developer Y contribute to release Z?”

The presentation describes the concepts of the Provenance data model and a software infrastructure for recording Provenance of software development processes. We show how the Provenance information helps to increase the quality of software releases and helps to give deep insight into the development process.