Metaherramientas MDE para el diseño de entornos de desarrollo de sistemas distribuidos de tiempo real

Laburpena

The main goal of this Thesis is the development of strategies and solutions for encouraging the adoption of Model-Driven Software Engineering (MDSE) by the experts in charge of designing development environments for software systems in general, and for distributed real-time systems in particular, thus fostering its implantation as the basis of such environments. In order to consolidate the adopted approach, the binomial software systems developer (environment user) and development environments designer is complemented by the definition of a third agent, named MDSE infrastructure developer. His main task is to alleviate the environments designer responsibility, so that he does not need expertise in MDSE technologies. Thus, the accomplishment of the main goal of the Thesis is addressed from the perspective of making contributions to the field of this third agent. A methodology has been designed for the development of generic tools that do not require adaptation when the meta-models constituting the conceptual scope of an MDSE-based environment evolve. The methodology envisages the functionality of each tool as depending on a model that instructs it in how to adapt itself to each target domain. More specifically, a generative approach has been adopted through which those generic tools are based on meta-tools that generate on demand the specific tool according to the specification formulated through the instructor model. The presented meta-tools are developed applying the Higher Order Transformation (HOT) technique for the automatic generation of model transformations. Under the same main goal of easing the environments specification and implementation tasks to the software engineering who is expert in their design, the Thesis proposes a generic conception of MDSE-based environments. The proposal considers that the environment user accomplishes its work through the supervised execution of processes, which consist in turn in the sequential and iterative execution of more basic operations, called tasks. According to the MDSE spirit, both the processes and the tasks are formulated and provided as models, being the environment the responsible of interpreting and executing them, always under the operator supervision. Several proof of concept implementations of the proposed strategies, methodologies and tools have been accomplished targeting the MAST-2 environment for the design and analysis of real-time systems.