VISUALIZATION OF CODES AND CODE SMELLS WITH 3D MUSHROOM FARM METAPHOR

• Main Author: Medianty Nababan, Anandhini
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/36854
• Institution: Institut Teknologi Bandung

Visualization for code comprehension is increasingly important as the size and complexity of the code increases. Various techniques and metaphors are used to visualize the code. In general, previous research on code visualization visualized code elements, relationships between code elements and metrics to indicate code smells. However, there is never a fundamental reason for why a specific property was chosen to be included in code visualization. In addition, observing the metric to find code smells is a difficult job. The 3D mushroom farm metaphor is a new metaphor that visualizes key object-oriented paradigm ideas, as properties that are important for understanding the code. 3D mushroom farm is a natural environment that is familiar in everyday life. Farms that must be managed properly represent codes that must also be managed properly. Mushrooms are chosen because of their simple structure, can provide clear visualization. Code properties are mapped to visual properties by observing good visualization characteristics. In contrast to other metaphors, the mushroom farm metaphor shows code smells directly through detection strategies. In addition, customization of detection strategies is possible for users because of the problem of human conceptualization about code smells. To describe the overall quality of the code, the percentage of the 'free-smell' code is displayed in numeric values. The proposed visualization design is implemented in a prototype to facilitate evaluation. The evaluation aims to test the functionality, effectiveness, usability, and usefulness of visualization. Broadly speaking, participants understood the purpose of visualization and agreed that code visualization with the mushroom farm metaphor was intuitive and able to represent code properties appropriately. Participants also managed to identify all elements affected by the code smells. The results of the study suggest code smell identification through detection strategies.