Detecting Design Violations in Django-based Web Applications

• Renieri Correia IFRN
• Eiji Adachi UFRN

Resumo

If on one hand frameworks allow programmers to reuse well-known architectural solutions, on the other hand they can make programmers unaware of important design decisions that should be followed during software construction, maintenance and evolution. And if programmers are unaware of these design decisions, there is a high risk of introducing design violations in the source code, and the accumulation of these violations might hinder software maintainability and evolvability. The use of static analysis tools might be employed to mitigate these problems by assisting the detection of recurring design violations in a given architectural pattern. In this work, we present MTV-Checker, a tool to assist the automatic detection of 5 design violations in Django-based web applications. We also conducted an empirical study in the context of the SUAP system, a large-scale Django-based information system with more than 175.000 lines of Python code currently deployed in more than 30 Brazilian institutions. Our results present the most recurrent violations, how they evolve along software evolution, and the opinions and experiences of software architects regarding these violations.

Palavras-chave: Code Smells, Design Violations, Python Applications, Software Architecture, Software Design, Web-Applications