Abstract
The modern software development landscape has seen a shift in focus toward mobile applications as smartphones and tablets near ubiquitous adoption. Due to this trend, the complexity of these "apps" has been increasing, making development and maintenance challenging. Current bug tracking systems do not effectively facilitate the creation of bug reports with useful information that will directly lead to a bug's resolution. To address the need for an improved reporting system, we introduce a novel solution, called Fusion, that helps reporters auto-complete reproduction steps in bug reports for mobile apps by taking advantage of their GUI-centric nature. Fusion links information, that reporters provide, to program artifacts extracted through static and dynamic analysis performed beforehand. This allows our system to facilitate the reporting process for developers and testers, while generating more reproducible bug reports with immediately actionable information.
Introduction
Software maintenance remains one of the most resource-intensive phases of the software lifecycle, with bug resolution at its core. However, a persistent "lexical knowledge gap" exists between the users or testers who discover bugs and the developers tasked with fixing them. Traditional bug trackers rely heavily on unstructured natural language, often resulting in vague descriptions that omit critical "steps to reproduce." This deficiency leads to a high volume of "non-reproducible" reports, which significantly drain developer productivity. While some crash-reporting tools exist, they often require heavy instrumentation and are limited to fatal crashes rather than logic or UI bugs.
The Fusion System Architecture
To bridge this gap, Fusion introduces a two-phase approach that leverages the GUI-centric nature of mobile applications to create a shared vocabulary between reporters and developers.
Phase 1: Analysis
The system begins with a pre-processing stage divided into two analysis types:
- Static Analysis (The Primer): Fusion utilizes tools such as
apktool,dex2jar, andsrcMLto extract GUI components (e.g., buttons, menus) from the application's source code or APK. This builds a comprehensive map of possible actions and links them to specific code-level artifacts. - Dynamic Analysis (The Engine): The system systematically explores the app using a Depth-First Search (DFS) algorithm. By employing
UIAutomator, it captures screenshots, component locations, and event flows to understand how different screens interact.
Phase 2: Report Generation
During the reporting process, Fusion acts as an intelligent assistant:
- Auto-completion Engine: As a reporter documents a bug, Fusion suggests the next likely steps based on the pre-built GUI model. Reporters select from suggested
{action, gui-component}tuples rather than typing free-form text. - Contextualized Reporting: For every step selected, Fusion automatically attaches a screenshot and a traceability link to the relevant source code, ensuring the developer has immediate, actionable context.
Experimental Evaluation
The researchers evaluated Fusion using 15 real-world bugs identified in 14 open-source Android applications from the F-Droid marketplace. The evaluation included two primary user studies:
- Report Creation: Eight graduate students generated 120 bug reports, split evenly between Fusion and the standard Google Code Issue Tracker (GCIT).
- Reproduction: Twenty participants then attempted to reproduce the bugs using the reports generated in the first study.
The results demonstrated a marked improvement in report quality. Only 10.8% of Fusion-generated reports were found to be non-reproducible, a significant decrease compared to the 19.1% non-reproducibility rate of reports created with the standard tracker.
Discussion and Key Insights
Fusion’s success lies in its ability to treat the GUI as a common language. By modeling the interface, the system guides the reporter to be precise, effectively reducing the cognitive load required to document complex reproduction steps. Unlike reactive tools that record failures after they occur, Fusion is proactive, facilitating the creation of high-quality data from the outset.
However, the system does face challenges. The initial static and dynamic analysis introduces overhead that must be repeated for every new version of an app. Furthermore, for extremely complex applications, the DFS "ripping" process may struggle to achieve full state coverage. Future work may involve leveraging unique mobile hardware features, such as sensors and touch gestures, to further streamline the reporting process.
Conclusion
Fusion represents a significant step forward in mobile bug reporting. By automating the capture of reproduction steps and linking them directly to source code, it reduces the communication gap between testers and developers, leading to faster resolution times and more robust mobile software.
Summary updated