What do practitioners expect from the meta-modeling tools? A survey

Abstract

Modeling languages are defined with a meta-model, which are specified using the meta-modeling tools that produce the editors for specifying models in accordance with the meta-models. In this paper, we aim to understand the top-used meta-modeling tools and practitioners’ expectations and challenges with respect to different requirements. So, we designed and conducted a survey, which was responded by 103 practitioners. The questionnaire considers the notation, semantics, editor services, model-transformation, validation, testing, and composability requirements. Some of the results are as follows: the top-used meta-modeling tools are Sirius and GEMS, Metaedit+, Xtext, and Microsoft DSL tools respectively. The top-preferred language visualizations are diagrammatic or textual. Practitioners prefer the editors with free-editing mode and are not familiar with projectional editing that promotes hybrid modeling. Translational semantics definition (i.e., model-to-model and model-to-text) is more popular than the interpretative semantics definition. Importing/exporting models and meta-models is highly important for facilitating the communication and collaboration. Modeling and meta-modeling with re-use and their versioning are among the top-desired features. Practitioners are willing to integrate any validation tools (e.g., model checkers) to define the languages’ semantical and structural validation rules and prove the model correctness. Lastly, defining the language semantics by composition and testing the semantics definitions are crucial for many practitioners. The survey also reveals many important challenges in each type of requirements. The survey results will be useful for meta-modelers in analyzing the meta-modeling tools and tool vendors in learning practitioners’ expectations.