ReMINDER: Uma Abordagem para Modelagem de Propriedades Não-Funcionais em Linhas de Produto de Software Dinâmicas

Abstract

Software Product Lines (SPL) represent a set of software systems that share a set of common features and managed to satisfy the needs of a particular market segment or mission. However, the representation of variability of products generated by SPL is made only of static way, that is, the adaptation occurs at design time. However, with the advent of context-sensitive systems, the development of systems has become more complex, given the changes of requirements and restrictions according to the environment to which it is inserted, requiring a high degree of adaptation.These systems can be conceptualized as Dynamic Software Product Lines (DSPL). One of the major challenges to build a DSPL is to develop a mechanism to incorporate scenarios with context adaptation and non-functional properties (NFP). To incorporate these mechanisms into the DSPL features model, a good strategy is to use an approach to guide the identification and representation of these mechanisms. In this scenario, the objective of this work is to represent NFPs and context adaptations scenarios in DSPL feature models. For this, an approach was developed to identify NFPs, constraints and context adaptations scenarios, activation and deactivation of features. The approach is called ReMINDER. An extension of the DyMMer tool was also developed to support the approach. Two studies were carried out to evaluate this work.The first one was a controlled experiment, to analyze the process defined in the ReMINDER approach, with the purpose of characterizing its application, in relation to the identification of NFPs and context adaptation scenarios, with their respective constraints, to support feature modeling and representation of constraints in DSPL feature models. The second study was the application of a usability test. This test was conducted to evaluate the usability of the tool extend through the analysis of four factors: (i) general satisfaction; ii) usability of the tool; (iii) the quality of information; and (iv) the quality of the interface. In general, after analysis and interpretation of all results obtained from studies, it was concluded that the approach has assisted in the identification of NFPs and its relations with the behavior of the feature models through interdependencies constraints and that the extended tool has good usability and is useful for operationalizing the approach.