Controlling Metamodel Availability Through a Maturity Model
| Vortragende(r) | Tobin Zühlke | |
|---|---|---|
| Vortragstyp | Masterarbeit | |
| Betreuer(in) | Erik Burger | |
| Termin | Fr 21. November 2025, 13:00 (Raum 010 (Gebäude 50.34)) | |
| Vortragssprache | Deutsch | |
| Vortragsmodus | in Präsenz | |
| Kurzfassung | Collaborative metamodel development often suffers from delays or premature exposure of
unstable parts of the metamodel due to the lack of mechanisms for controlling the availability of metamodel changes. In practice, this leads to two major issues: either metamodel changes must wait on each other before being delivered, or changes are included in the delivery although they are not yet suitable for certain stakeholders. To address these problems, we introduce the Maturity-Management-System (MMS), which leverages a descriptive maturity model to classify metamodel changes into distinct stages, ranging from Initial to Released. By integrating this model with the Transitional Metamodel approach—a metamodeling technique that preserves old states for each change—the MMS de- rives the maturity of individual metamodel elements from the maturity of the corresponding metamodel changes. Based on the maturity assessment of individual metamodel elements, the approach enables filtering of immature parts of the metamodel during code generation and disables metamodel elements at runtime depending on the maturity requirements of the end users. To ensure metamodel consistency while allowing filtering and disabling of parts of the metamodel, a set of metamodel invariants is defined. The MMS enforces these invariants through the Reject and Synchronize strategies. Using the metamodel of the model-based electric/electronic development environment PREEvision as an example, it was demonstrated that 64 % of MOF-compliant metamodel changes are supported by the prototype. Evaluation results further show that the approach significantly reduces the waiting time between the completion of a metamodel change and its availability for code generation—by up to 88 %—while introducing only a marginal increase in metamodel complexity. | |