New graduates joining the software engineering workforce sometimes have trouble writing test code. Coming from university, they lack a hands-on approach to testing and have little experience with writing tests in a real-world setting. Software companies resort to costly training camps or mentoring initiatives. Not overcoming this lack of testing skills early on can hinder the newcomer’s professional progress in becoming a high-quality engineer. Studying open source developers, we found that they rely on a project’s pre-existing test code to learn how to write tests and adapt test code for their own use. We propose to strategically present useful and contextual test code examples from a project’s test suite to newcomers in order to facilitate learning and test writing. With an automatic suggestion mechanism for valuable test code, the newcomer is enabled to learn how senior developers write tests and copy it. Having access to suitable tests lowers the barrier for writing new tests.

Inexperienced software developers - for example, undergraduates entering the workforce - exhibit a lack of testing skills. They have trouble understanding and applying basic testing techniques. These inexperienced developers are hired by software companies, where this lack of testing skills has already been recognized. Companies allocate valuable resources and invest time and money in different onboarding strategies to introduce new hires to the organization’s testing practices. However, if the lack of testing skills is not addressed properly, the new hire is left to her own devices. This hinders her in becoming a high-quality engineer for the software company. This thesis proposes to improve the onboarding strategies with traits of social transparency in order to specifically address testing issues of inexperienced new hires. Social transparency has been shown to influence the testing behavior of development teams on a social coding site. An environment that is open for discussion helps newcomers to understand and adapt a team’s testing culture. Tailoring the onboarding process to better address testing skills of new hires makes it more effective and more efficient. This reduces the danger of carrying new hire’s testing deficits into commercial software development.

There are many different approaches to testing software, with different benefits for software quality and the development process. Yet, it is not well understood what developers struggle with when getting started with testing - and why some do not test at all or not as much as would be good for their project. This missing understanding keeps us from improving processes and tools to help novices adopt proper testing practices. We conducted a qualitative study with 97 computer science students. Through interviews, we explored their experiences and attitudes regarding testing in a collaborative software project. We found enabling and inhibiting factors for testing activities, the different testing strategies they used, and novices’ perceptions and attitudes of testing. Students push test automation to the end of the project, thus robbing themselves from the advantages of having a test suite during implementation. Students were not convinced of the return of investment in automated tests and opted for laborious manual tests - which they often regretted in the end. Understanding such challenges and opportunities novices face when confronted with adopting testing can help us improve testing processes, company policies, and tools. Our findings provide recommendations that can enable organizations to facilitate the adoption of testing practices by their members.

Many software development projects struggle with creating and communicating a testing culture that is appropriate for the project's needs. This may degrade software quality by leaving defects undiscovered. Previous research suggests that social coding sites such as GitHub provide a collaborative environment with a high degree of social transparency. This makes developers' actions and interactions more visible and traceable. We conducted interviews with 33 active users of GitHub to investigate how the increased transparency found on GitHub influences developers' testing behaviors. Subsequently, we validated our findings with an online questionnaire that was answered by 569 members of GitHub. We found several strategies that software developers and managers can use to positively influence the testing behavior in their projects. However, project owners on GitHub may not be aware of them. We report on the challenges and risks caused by this and suggest guidelines for promoting a sustainable testing culture in software development projects.

GitHub projects attract contributions from a community of users with varying coding and quality assurance skills. Developers on GitHub feel a need for automated tests and rely on test suites for regression testing and continuous integration. However, project owners report to often struggle with implementing an exhaustive test suite. Convincing contributors to provide automated test cases remains a challenge. The absence of an adequate test suite or using tests of low quality can degrade the quality of the software product. We present an approach for reducing the effort required by project owners for extending their test suites. We aim to utilize the phenomenon of drive-by commits: capable users quickly and easily solve problems in others' projects - even though they are not particularly involved in that project - and move on. By analyzing and directing the drive-by commit phenomenon, we hope to use crowdsourcing to improve projects' quality assurance efforts. Valuable test cases and maintenance tasks would be completed by capable users, giving core developers more resources to work on the more complicated issues.