Sporathe knowledge network
ES
Universidad Nacional de Colombia

Medellín - Minas - Maestría en Ingeniería - Ingeniería de Sistemas · 2021

A method to identify toggle debt in a project from its source code.

Continuous Delivery has benefited companies such as IBM, Google and Microsoft after the adoption of practices to deliver friction-less value. These lean and agile practices help teams to adjust to implicit variability when developing software and to provide non-stop features and enhancements. Practitioners that use feature toggles in their source code can control the execution of features with a high degree of precision. As this execution control effectively decouples the feature implementation from its release, feature toggles help practitioners move valuable software artifacts closer to the user. However, there is evidence of unwanted risks and negative consequences that impact the quality of a software project when, for example, a feature hidden under a toggle remains in the source code longer than expected. Although practitioners of feature toggles have crafted mechanisms to keep their feature toggles inventory in check, little is known about the automated identification of feature toggles and their associated technical debt. This thesis focuses on the proposal of a method to identify source code where both feature toggles and technical debt coincide in the repository of a project. This method is based on two complementary empirical explorations. First, we identified areas of code containing feature toggles in 72 open source projects that use toggling libraries. And second, we selected one of the toggled projects and identified technical debt that intersects with the previously identified feature toggles. The proposed method uncovers the presence of toggle debt in the early stages of the development process. These results demonstrate the feasibility of a systematic approach to find toggle debt from a software repository, and advise practitioners on the importance for automated, reliable and up-to-date quality measurement tools.

Full text 67 pages with text out of 73

Read the full thesis Record in the repository

Its library's thesaurus