Comparing relational model transformation technologies: implementing Query/View/Transformation with Triple Graph Grammars

Abstract

The Model Driven Architecture (MDA) is an approach to develop software based on different models. There are separate models for the business logic and for platform specific details. Moreover, code can be generated automatically from these models. This makes transformations a core technology for MDA and for model-based software engineering approaches in general. Query/View/ Transformation (QVT) is the transformation technology recently proposed for this purpose by the OMG. Triple Graph Grammars (TGGs) are another transformation technology proposed in the mid-nineties, used for example in the FUJ-ABA CASE tool. In contrast to many other transformation technologies, both QVT and TGGs declaratively define the relation between two models. With this definition, a transformation engine can execute a transformation in either direction and, based on the same definition, can also propagate changes from one model to the other. In this paper, we compare the concepts of the declarative languages of QVT and TGGs. It turns out that TGGs and declarative QVT have many concepts in common. In fact, QVT-Core can be mapped to TGGs. We show that QVT-Core can be implemented by transforming QVT-Core mappings to TGG rules, which can then be executed by a TGG transformation engine that performs the actual QVT transformation. Furthermore, we discuss an approach for mapping QVT-Relations to TGGs. Based on the semantics of TGGs, we clarify semantic gaps that we identified in the declarative languages of QVT and, furthermore, we show how TGGs can benefit from the concepts of QVT.Keywords MDA · Model-based software engineering · Model transformation · Query/View/Transformation (QVT) · Triple Graph Grammar (TGG)

Controlling Reuse in Pattern-Based Model-to-Model Transformations

“…However, this is still an open question. We would also like to explore higher-level means of specifications, by (i) omitting the correspondence graph at the specification level (and automatically generating the traces at the operational level, as in [10,14]), and (ii) making possible the specification of pattern dependencies and parameter passing, similar to when or where clauses in QVT. These two steps would allow us to express the semantics of QVT-R with our framework.…”

“…In our case, we assume that not every element in the source needs to be translated, but the fact that we generate several rules for each pattern permits obtaining all valid target models, if more than one exists [21]. An attempt to bridge TGGs and QVT-R is [10], where QVT-R is both compiled into operational TGGs (instead of using QVT-core) and translated into declarative TGGs.…”

Correctness of Model Synchronization Based on Triple Graph Grammars

“…Triple graph grammars have been successfully applied in multiple case studies for bidirectional model transformation, model integration and synchronization [17,21,9,8], and in the implementation of QVT [12]. Moreover, several formal results are available concerning correctness, completeness, termination [7,10], functional behavior [16,10], and optimization with respect to the efficiency of their execution [14,18,10].…”

“…Triple graph grammars (TGGs) have been successfully applied in several case studies for bidirectional model transformation, model integration and synchronization [17,21,9,8], and in the implementation of QVT [12]. Inspired by Schürr et al [20,21], we started to develop a formal theory of TGGs [7,14], which allows us to handle correctness, completeness, termination, and functional behavior of model transformations.…”

Toward Bridging the Gap between Formal Semantics and Implementation of Triple Graph Grammars

“…[13]). For TGGs, there are different tools, which realize slightly different dialects, such as Fujaba TGG Engine [3], MOFLON [1], or ATOM3 [19].…”