Fine-grained Analysis of Global Software Development Process

This paper proposes a method for micro process analysis of ongoing distributed software development. Micro process analysis firstly identifies process instances by matching logs of software development tools with pre-defined micro (fine-grained) process models. Micro process metrics can be measured from identified process instances. In distributed or multisited software development environment, micro process metrics provide project manager with more visibility to precise process execution and also to precise monitoring. As an empirical study, we retrospectively applied micro process analysis to bug fixing process in organizations attending a multi-site project of commercial software development. In the study, process instances are extracted from histories of CVS (a source code configuration management system) and GNATS (a bug tracking system) based on a bug fixing procedure predefined in the project. We confirmed that extracting and visualizing process instances helps to identify accuracy of process executions as well as accuracy of logging.