Model-driven architecture view consistency checking
Ekşi, Gülsüm Ece
MetadataShow full item record
Please cite this item using this persistent URLhttp://hdl.handle.net/11693/28925
Software architecture is one of the key artefacts in the software development process since it provides the gross-level structure of the system and supports the stakeholder concerns. To benefi t from the architecture it is important that the code is consistent with the architecture and the corresponding design decisions. Unfortunately, changing requirements and/or the adaptations to the code can lead to undesired inconsistencies among the architecture and the code. This so-called architectural drift problem is the discrepancy between the architecture description and the resulting implementation. Several approaches have been proposed to detect the inconsistencies between the software architecture and the code to ensure that the original design goals are maintained. In practice, software architecture is documented using a coherent set of architecture views, each of view addresses particular stakeholder concerns. Similar to the consistency with the code it is important that an architecture view is consistent within itself and with other related architecture views. Unfortunately, the existing architecture conformance analysis approaches have primarily focused on checking the inconsistencies between the architecture and code, and did not explicitly consider the consistency among views. In this thesis, we provide a systematic architecture conformance analysis approach that explicitly focuses on conformance analysis among architecture views. The approach is used for detecting the inconsistencies within and across architectural views. To this end, we define the meta-models of architecture viewpoints, present the conformance analysis approach, and provide the tool ArchViewChecker. We illustrate our approach for detecting inconsistencies using the Views and Beyond approach. We adopt a fault injection approach to evaluate the effectiveness of the approach. The results show that the approach is effective in detecting inconsistencies within views and across views.