| APR (Automated Program Repair) |
Techniques (search-based, learning-based, etc.) used to identify and fix bugs in software code through automated means. |
| Assertion Roulette |
A test smell involving multiple assertion statements in a single test case without descriptions, making it hard to identify which specific assertion failed. |
| Chain of Thought (CoT) |
A prompting strategy that encourages the model to generate a sequence of intermediate reasoning steps before providing the final answer. |
| Code Smells |
Surface indications in code that usually correspond to deeper quality issues, such as poor readability or maintainability, without necessarily being functional bugs. |
| Context Engineering |
The practice of enriching prompts with additional information, such as API documentation, function signatures, or error messages, to improve model output. |
| Few-shot Learning |
Providing a small number of examples within a prompt to help the LLM understand the desired pattern or task format. |
| Functional Bug |
An error that causes a program to fail to meet its intended requirements or produce incorrect results despite potentially being syntactically correct. |
| Hallucination |
A phenomenon where an LLM generates plausible-sounding but factually incorrect or non-existent code elements, such as fake libraries or parameters. |
| Logic Bug |
A subcategory of functional bugs where the algorithm or business logic is flawed, leading to incorrect behavior during execution. |
| Magic Number |
A test smell where specific, hard-coded values are used in assertions without explanation or being assigned to named constants. |
| Memory Bug |
A system-level defect involving improper management of computer memory, such as out-of-bounds reads/writes or infinite recursion leading to stack overflow. |
| RAG (Retrieval-Augmented Generation) |
A technique that enriches prompts by retrieving relevant external data (e.g., similar code snippets) to provide more context to the model. |
| Reliability Bug |
Issues related to the software's ability to operate without failure, typically manifesting as performance bottlenecks or stability issues like deadlocks. |
| Semantic Bug |
An error where the code is syntactically valid, but the meaning or intention is incorrectly expressed, leading to unintended outcomes. |
| SLR (Systematic Literature Review) |
A rigorous research method used to identify, analyze, and synthesize all available empirical evidence related to a specific research question. |
| Syntax Bug |
Code that violates the grammatical rules of a programming language, preventing it from being compiled or interpreted. |
| Zero-shot Learning |
A scenario where a model is asked to perform a task without being provided any prior examples in the prompt. |
Top comments (0)