Browsing by Autor "Larissa Rocha"

Now showing 1 - 4 of 4

CRAFTPy: allowing people with visual impairments to create diagrams
(2024) Lucas Lopes Fraga; Ricardo Santos; Larissa Rocha
The inclusion of students with visual impairments (SVI) in higher education has greatly advanced with the advent of newtechnologies. Despite these strides, challenges remain, particularly in Information and Communication Technology (ICT) courses that heavily rely on visual elements. Screen readers facilitate SVI’s access to digital content, but many educational tools are still incompatible with these readers. This incompatibility is especially problematic in tools requiring interaction with visual components, such as block-based programming, diagramming, and 3D modeling tools. This study introduces CraftPy, an accessible web tool fully compatible with screen readers. CraftPy enables SVI to create various types of diagrams using Python code, employing an object-oriented approach to design classes, actors, entities, attributes, and relationships. We also conducted a preliminary evaluation involving eight SVI participants to assess the tool’s effectiveness. Overall, participants found the tool to be highly accessible with screen readers and user-friendly. They were able to complete the experiment tasks with minimal difficulties. However, improvements are needed, particularly in enhancing screen responsiveness for low-vision users who depend on screen magnifiers. By developing CraftPy, we aim to promote equity in higher education, offering SVI enhanced opportunities to succeed in ICT courses. Link to the video: https://youtu.be/NXu4xbOH8Q4
Identifying and Addressing Test Smells in JavaScript: A Developer-Centric Study
(2025) Jamille Carmo Oliveira; Luigi Mateus; Gabriel Amaral; Tássio Virgínio; Carla Bezerra; Ivan Machado; Larissa Rocha
Test smells are poor practices in test code that can compromise maintainability, reliability, and clarity. While the concept has been widely studied in languages such as Java and Python, research on test smells in JavaScript remains limited—despite its prominence in modern development. To address this gap, we conducted a focus group study with JavaScript developers of varying experience levels to explore their perceptions of seven test smells. These smells—Anonymous Test, Comments Only Test, Overcommented Test, General Fixture, Test Without Description, Transcripting Test, and Sensitive Equality—are particularly relevant to the JavaScript ecosystem and had not been systematically examined in this context prior to our study. We applied thematic analysis to transcribed discussions, uncovering developers’ concerns, recognition patterns, and proposed mitigation strategies. Our results show that experience level strongly influences the ability to detect and refactor test smells, with junior developers often struggling to identify more subtle patterns. To the best of our knowledge, this is the first study to investigate JavaScript developers’ perceptions of test smells using a qualitative approach. Our findings reveal key challenges, offer practical insights for test improvement, and support the development of better training and tooling for JavaScript test quality.
Improving JavaScript Test Quality with Large Language Models: Lessons from Test Smell Refactoring
(2025) Gabriel Amaral; Henrique L. Gomes; Eduardo Figueiredo; Carla Bezerra; Larissa Rocha
Test smells—poor design choices in test code—can hinder test maintainability, clarity, and reliability. Prior studies have proposed rulebased detection tools and manual refactoring strategies, most focus on statically typed languages such as Java. In this paper, we investigate the potential of Large Language Models (LLMs) to automatically refactor test smells in JavaScript, a dynamically typed and widely used language with limited prior research in this area. We conducted an empirical study using GitHub Copilot Chat and Amazon CodeWhisperer to refactor 148 test smell instances across 10 real-world JavaScript projects. Our evaluation assessed smell removal effectiveness, behavioral preservation, introduction of new smells, and structural code quality based on six software metrics. Results show that Copilot removed 58.78% of the smells successfully, outperforming Whisperer’s 47.30%, while both tools preserved test behavior in most cases. However, both also introduced new smells, highlighting current limitations. Our findings reveal the strengths and trade-offs of LLM-based refactoring and provide insights for building more reliable and smell-aware testing tools for JavaScript.
SNUTS.js: Sniffing Nasty Unit Test Smells in Javascript
(2024) Jamille Carmo Oliveira; Luigi Mateus; Tássio Virgínio; Larissa Rocha
Test smells indicate potential issues or weaknesses within the test code, which can compromise its effectiveness and maintainability. They highlight areas where improvements can enhance the overall quality of the test suite or testing practices. For instance, an example of a test smell is the Anonymous Test, where the test’s name lacks descriptive information about its function or purpose. Addressing these test smells can result in more robust and maintainable test suites, thus improving the reliability of the testing process. Despite significant research on these issues, tools are scarce for automatically detecting them, particularly in certain programming languages such as JavaScript. In the current landscape, existing test smell detection tools for JavaScript lack intuitiveness and graphical interfaces, and require extensive configuration, which may lead to low adoption within the developer community. To address this gap, we propose SNUTS.js, a tool designed to streamline the detection of test smells in JavaScript. Designed as an API, SNUTS.js offers versatility, allowing integration with various tools and environments. This tool goes beyond existing solutions by identifying previously undetected test smells, including the Anonymous Test, Comments Only Test, Overcommented, General Fixture, Transcripting Test, and Sensitive Equality. We also introduce a new test smell termed Test Without Description, which denotes a test case lacking descriptive text. In a preliminary evaluation, we constructed a dataset of tests sourced from real-world projects on GitHub. Through manual analysis, we identified 285 instances of test smells. SNUTS.js demonstrated a detection accuracy of 100% for three specific types of test smells, Anonymous Test, Overcommented, and General Fixture, all tailored to the JavaScript environment. Link to the video: https://youtu.be/89z0jy4Nu0s