Um estudo sobre as práticas de testes funcionais de GUI na indústria brasileira de software

Abstract

Software testing is an important activity in the software development cycle, whichseeks to ensure the quality of software created. When software does not meet itsusers’ implicit and explicit needs, one can say that it lacks quality. The lack of qualityin software can generate several problems for the software development project, suchas friction between supplier and clients/users, financial losses, legal problems due tobreach of contracts, and even risk to the life of its users in critical contexts.Along these lines, different levels of testing have been developed to ensure quality indifferent aspects of the software. A popular approach is functional interface testing,which evaluates software by executing events on the software screen, such as fillingin a text field or clicking in a button. GUI (Graphical User Interface) functional testing can be executed either manually, with the tester interacting directly with the software, or automatically, by running the software through a programmed script. In recent years, one can find several solutions and techniques proposed to deal with GUI testing in the literature. However, they usually do not analyze how industry professionals perform this type of testing, leaving a gap in the area. Therefore, this study investigates how testers perform interface testing on their projects. First, an online survey was conducted with 222 testing professionals with different backgrounds, job titles, and roles. The results indicated that many professionals perform tests exclusively in a manual way. Additionally, most of them agree that tools used in test automation meet business needs, although they have technical limitations. Then, we conducted a semi-structured interview with 20 testing professionals to complement the results obtained with the questionnaire. The results indicated that testers who only perform manual GUI tests do not create automated GUI tests due to the context of the project in which they work, their professional preferences and also for educational reasons, such as difficulty in programming. In addition, distinct limitations were observed for manual and automated GUI tests that negatively impact the activities performed by professionals, such as difficulties related to the company’s culture and technical problems related to the software under test. The combined results provide a general overview of GUI testing available in the Brazilian software industry.