Sourceminer trends: uma infraestrutura para análise de evolução de software baseada em múltiplas visões

Abstract

Analyzing the dynamics related to software evolution allow users and stakeholders to adopt measures to keep a program useful. Software evolution visualization tools are one of the research lines towards this direction. They have the potential to aggregate large amounts of data, as well as allowing comparisons, highlighting outliers, among other aspects, in an easy-to-read interface. However, the heterogeneity of evolutionary data sources as well as the correlation among them is challenging. At the same time, the available visualization tools tend to focus on data representation rather than focusing on meeting certain comprehension goals that a user might have. In this work, Sourceminer Trends is presented, an evolution analysis tool that seeks to solve these problems in an objective-oriented approach. Models are presented to characterize data evolution, source code and visual structures, as well as the mapping between them. These models, as well as the transformation process is presented through use examples. It was observed that the adoption of models make the analysis environment more flexible and facilitate the serialization of the retrieved data. As limitations, it is worth mentioning that the use of models has led to additional performance problems. In addition, the mapping between objectives and visual representations presented a greater level of coupling than expected.