Abstract
The modern software development landscape has seen a shift in focus toward mobile applications as tablets and smartphones near ubiquitous adoption. Due to this trend, the complexity of these apps has been increasing, making development and maintenance challenging. Additionally, current bug tracking systems are not able to effectively support construction of reports with actionable information that 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 users auto complete reproduction steps in bug reports for mobile apps. FUSION links user provided information to program artifacts extracted through static and dynamic analysis performed before testing or release. The approach that FUSION employs is generalizable to other current mobile software platforms, and constitutes a new method by which off device bug reporting can be conducted for mobile software projects. In a study involving 28 participants we applied FUSION to support the maintenance tasks of reporting and reproducing defects from 15 real world bugs found in 14 open source Android apps while qualitatively and qualitatively measuring the user experience of the system. Our results demonstrate that FUSION both effectively facilitates reporting and allows for more reliable reproduction of bugs from reports compared to traditional issue tracking systems by presenting more detailed contextual app information.
Introduction
A fundamental disconnect exists in software maintenance: bug reporters (users or testers) typically possess functional knowledge of an application, while developers have intimate code-level knowledge. This "lexical gap" often results in bug reports written in vague natural language, lacking the specific steps or technical context required for a fix. In the mobile ecosystem, this problem is exacerbated by complex, event-driven Graphical User Interfaces (GUIs). Traditional issue trackers like Bugzilla or GitHub rely on unstructured text, leading to a high rate of non-reproducible bugs. Research suggests that only about 50% of developers find bug reports complete, with providing reproduction steps cited as the most difficult task for reporters.
The FUSION Framework
To bridge this gap, we developed FUSION, a framework that operates on an Analyze → Generate paradigm. FUSION automates the collection of application metadata to assist users in constructing precise reproduction steps.
Analysis Phase
Before an application is released or tested, FUSION performs a two-part analysis to map its internal structure and behavior:
- Static Analysis (The "Primer"): Using tools such as
apktoolanddex2jar, FUSION extracts GUI components, their types (e.g., buttons, spinners), and the specific Java classes or activities where they reside. - Dynamic Analysis (The "Engine"): FUSION employs a systematic depth-first search (DFS) using the
UIAutomatorframework. It "rips" the application to discover possible state transitions, capturing screenshots and precise screen coordinates for every interactive element.
Report Generation Phase
When a user encounters a bug, they use FUSION’s web-based portal to report it. As the reporter begins typing a reproduction step, FUSION’s Auto-Completion Engine queries the pre-built database to suggest the most likely next actions and components based on the app’s current state model.
By forcing users to select from a list of known GUI components rather than typing free-form text, FUSION eliminates ambiguity. Even if a user does not know the technical name of a widget, they can identify it via the suggested screenshots provided in the auto-completion menu.
Evaluation and Results
We evaluated FUSION using 14 open-source Android apps from F-Droid (e.g., GnuCash, OI Notepad) and 15 real-world bugs of varying complexity. A user study with 28 participants—comprising both "expert" graduate students and "non-expert" undergraduates—compared FUSION against the Google Code Issue Tracker (GCIT).
Key Findings
- Improved Reproducibility: Reports generated with FUSION were significantly more reliable. Only 10.8% (13/120) of FUSION reports were non-reproducible, compared to 19.2% (23/120) for the traditional GCIT system.
- Bridging the Expertise Gap: FUSION effectively "leveled up" inexperienced reporters. Non-experts using FUSION produced reports that were nearly as easy to reproduce as those written by experts, as the structured workflow forced the inclusion of detailed technical information.
- Developer Preference: While creating FUSION reports took slightly longer due to the added detail, experienced developers overwhelmingly preferred them for their clarity and inclusion of actionable metadata, such as specific Java classes involved in the fault.
Conclusion and Future Work
FUSION demonstrates that linking user-provided information to program artifacts through automated analysis can significantly improve the quality of mobile bug reports. While current limitations include limited support for complex gestures and external sensor dependencies, future work will focus on expanding the dynamic engine and exploring automated fault localization using FUSION's structured data.
Summary updated