Resumen de A navigational role-centric model oriented web approach moweba
Some of the major challenges facing Web applications today are those of portability, adaptability and evolution, not only in the environment in which they run, but also in the way in which they must be developed, often requiring different languages, frameworks, tools, environments, platforms, etc. MDD and MDA take into account these issues. However, to achieve portability, adaptability and evolution depends to some extent on the degree of independence that the models adopt.
This Thesis presents a method that take into account the problem of evolution and portability towards different environments. The approach is called MoWebA (Model Oriented Web Approach). Some key aspects of MoWebA that could have a positive impact in the portability and adaptability are:i) incorporation of an Architecture Specific Model (ASM) as a new modeling layer, in order to keep the portability of the Platform Independent Model (PIM) regarding the different architectures (e.g., RIA, SOA, Mobile); ii) clear separation of the presentation layer with regard to the navigation and behavior layers; iii) definition of the navigational structure according to a function-oriented approach, which prevents the modification of the navigation design caused by implementation changes; iv) and use of standards in order to facilitate the independence from the tools.
We justify MoWebA by highlighting a series of concerns for Web applications development. We present an overview of the method including the dimensions and the diagrams that we propose. Subsequently, we present each step in the modeling process, including the diagrams and notation, its definition (metamodels) and examples of use. Afterwards, we present the transformation process adopted by MoWebA, which includes model-to-model and model-to-code transformations.
We have devoted special attention to the validation of the approach. As a first validation, MoWebA has been used for modelling and generating different types of applications by both novice and experienced modellers and developers. These experiences were done in academic and industrial contexts. The experiences have allowed to identify strengths and weaknesses of the PIM proposal, and to verify that the proposed notation covers the needs of different domains. Next, we present a preliminary validation of the ASM proposal, considering an experience of different ASM definitions made by a group of computer science students at the Catholic University "Nuestra Señora de la Asunción" (Paraguay). This preliminary validation has allowed us to determine how feasible is to adapt the proposal to other architectures. The analysis of the validation sought to answer the following questions: Can the same PIM model be used for different architectures?; Is it possible to specify clear limits between platform independent models (PIM) and architectural specific models (ASM)?; How does an architectural specific model facilitate the transformation rules definition?. Finally, we present a Case Study to validate the extensions of MoWebA to three different architectures. The experience was structured taking into account the framework of Runeson et al. The Case Study was done as part of a research project led by the Catholic University of Asunción called "Mejorando el Proceso de Desarrollo de Software: Una propuesta basada en MDD" (https://www.dei.uc.edu.py/proyectos/mddplus/), grant 14-INV-056 of CONACYT (Consejo Nacional de Ciencias y Tecnología), Paraguay. This experience have allowed to carry out three complete extensions. In such extensions we could analysed the grade of adaptability of MoWebA and of automation PIM-ASM, as well as the grade of independence of the PIM metamodel. We have also conducted some user's satisfaction experiences with modelers and developers.
