A Declarative Framework for Hand-Crafted Mutation Analysis and Management
New declarative system standardizes hand-crafted mutants, enabling selective execution and lossless conversion across representations.
Researcher Alperen Keles has introduced a comprehensive declarative framework for hand-crafted mutation analysis and management, addressing fragmentation in current tooling that forces trade-offs between readability, mutation preservation, and execution cost. The framework, implemented in a prototype system called Marauder, systematically characterizes five distinct mutation representations used in software testing: comment-based, preprocessor-based, patch-based, match-and-replace, and in-AST mutations. This classification provides clarity in a previously disorganized design space where researchers evaluating fuzzing and property-based testing tools for AI systems often struggled with inconsistent approaches.
Keles defines a formal mutation algebra that supports advanced operations including selective execution of specific mutants, tag-based expansion for grouping related mutations, and higher-order combinations of multiple mutants. The framework's core innovation is a lossless conversion pipeline that maps between different mutation representations through a common intermediate form, with particular attention to extracting and normalizing in-AST mutations. Marauder serves as a practical implementation, providing tools for injecting, activating, resetting, and composing hand-crafted mutants across all five representation types. This unified approach enables researchers to conduct more efficient and expressive mutation experiments while maintaining consistency across different testing methodologies.
- Characterizes five mutation representations: comment-based, preprocessor-based, patch-based, match-and-replace, and in-AST mutations
- Defines mutation algebra supporting selective execution, tag-based expansion, and higher-order mutant combinations
- Implements lossless conversion pipeline through Marauder prototype for unified mutation management across representations
Why It Matters
Provides standardized methodology for evaluating AI testing tools, enabling more reliable comparisons and reproducible research in software engineering.