A rigorous methodology for developing GUI-based DSL formal tools

• Ano de defesa: 2013
• Autor(a) principal: Silva, Robson dos Santos e
• Orientador(a): Mota, Alexandre Cabral
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal de Pernambuco
• Palavras-chave em Português: Engenharia Dirigida a Modelos (MDE); Métodos Formais; Linguagens específicas de domínio; Ferramentas formais com interface gráfica; Model-Driven Engineering; Formal Methods; Domain-Specific Languages; GUI-based formal tools
• Link de acesso: https://repositorio.ufpe.br/handle/123456789/12366

It is well-known that model-driven engineering (MDE) is a software development methodology that focuses on creating and exploiting (specific) domain models. Domain models (conceptually) capture all the topics (for instance, entities and their attributes, roles, and relationships as well as more specific constraints) related to a particular problem. It is common to use domain-specific languages (DSL) to describe the concrete elements of such models. MDE tools can easily build domain-specific languages (DSL), capturing syntactic as well as static semantic information. However, we still do not have a clear way of capturing the dynamic semantics of a DSL as well as checking the domain properties prior to generating the implementation code. Formal methods are a well-known solution for providing correct software, where we can guarantee the satisfaction of desired properties. Unfortunately the available formal methods tools focus almost exclusively on semantics whereas human-machine interaction is "left to the user". Several industries, and in particular the safety-critical industries, use mathematical representations to deal with their problem domains. Historically, such mathematical representations have a graphical appeal. For example, Markov chains and fault-trees are used in safety assessment processes to guarantee that airplanes, trains, and other safety-critical systems work within allowed safety margins. In general, due to the difficulty to obtain correct software, such industries use Commercial Off-The-Shelf (COTS) software or build them specifically to satisfy their needs with a related testing campaign effort. Such DSLs are difficult to capture, using just MDE tools for instance, because they have specific semantics to provide the desired (core) information for the industries that use them. In this sense, given a DSL (L) composed of a syntax and static semantics (SSL), and dynamic semantics (DSL) parts, our work proposes a rigorous methodology for combining the easiness of MDE tools, to capture SSL, with the correctness assured by formal methods, to capture DSL as well and check its properties. This combination is specifically handled in the following way, we capture all aspects of L using formal methods, check the desired properties and adjust if necessary. After that, we automatically translate part of it in terms of constructs of a MDE tool, from which we can build a user-friendly (GUI) front-end very easily (automatically). Finally, we link the front-end code to the automatically synthesized code from the formal dynamic semantics back-end. Although we require the use of a formal methods tool, the distance from the mathematical representations used in industry and the formal methods notation is very close. With this proposed methodology we intend that safety-critical industries create their domain specific software as easy as possible and with the desired static and dynamic properties formally checked.