Conflict Detection on Models An Empirical Study

Software Configuration Management (SCM) systems are employed to facilitate collaboration on software engineering artifacts such as models. Therefore, they need to detect concurrent change – also known as conflict detection. However, many researchers recognized that conflict detection on models is not well supported in traditional SCM systems since they are engineered for textual files. Their conflict detection shows many false positives, and they require a merge every time the same the same file is changed. Operation-based conflict detection and merging have been proposed to overcome these problems and reduce false positives. In this paper, we present an empirical study to compare traditional file-based, textual conflict detection with operation-based conflict detection on models. We compare two different strategies of file-based with two operation-based strategies in a multi-case study. We show that operationbased conflict detection results in fewer conflicts and thus requires fewer merges.