Code Reviews for Visual Programming Languages

• Main Author: Ragusa, Giuliano Giorgio
• Publication Date: 2018
• Format: Master thesis
• Language: eng
• Source: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Download full: http://hdl.handle.net/10362/75494
• topic: Code Review; Visual Programming Languages; Inspection processes; Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Eletrotécnica, Eletrónica e Informática

Visual languages such as OutSystems are being progressively more used by large teams that collaborate on enterprise grade projects. This results in new challenges when it comes to ensuring the quality of the applications built with this languages, without sacrificing productivity. An increasingly common way to improve quality is performing distributed lightweight tool supported code reviews. A code review consists of an analysis of the code by one or more developer. The objective for this is to ensure that the changes do not introduce bugs and adhere to the established coding best practices. With this practice, companies can offer higher quality and easier to maintain software. However, the current tools and processes for code review are focused on textual programming languages and opening a bigger gap for the visual ones. The main goal of this dissertation is to develop a specialized tool to promote and making visual programming languages code review more accessible. An easy-to-follow process was also created to support the tool. At the moment, the tool can be used with OutSystems’ language and UML class diagram, however, new languages can easily be added to the solution. We examined the code review method currently used by developers vs. the new method created concerning defect detection rates and user experience. Also conducted individual interviews with 30 computer science engineers. Results were collected with the help of questionnaires and analyzed using multiple descriptive statistics. The new code review method was significantly better than the current method used, reaching average usability improvements of 47.75%, tasks load reduction of 15,43%. Generalization of results is limited since the analyzed techniques were applied only to a small set, consisted of 30 testers. The result supported the evidence concerning the advantage of the new code review method using the developed tool over the current method of reviewing visual artifacts.