MUTAGEN: Reliable Coverage-Guided, Property-Based Testing using Exhaustive Mutations
Automatically-synthesized random data generators are an appealing option when using property-based testing. There exists a variety of techniques that extract static information from the codebase to produce random test cases. Unfortunately, such techniques cannot enforce the complex invariants often needed to test properties with sparse preconditions.
Coverage-guided, property-based testing (CGPT) tackles this limitation by enhancing synthesized generators with structure-preserving mutations guided by execution traces. Albeit effective, CGPT relies largely on randomness and exhibits poor scheduling, which can prevent bugs from being found.
We present MUTAGEN, a CGPT framework that tackles such limitations by generating mutants exhaustively. Our tool incorporates heuristics that help to minimize scalability issues as well as cover the search space in a principled manner. Our evaluation shows that MUTAGEN not only outperforms existing CGPT tools but also finds previously unknown bugs in real-world software.
Tue 18 AprDisplayed time zone: Dublin change
11:00 - 12:30 | Session 8: Mutation TestingJournal-First Papers / Research Papers at Grand canal Chair(s): Renzo Degiovanni SnT, University of Luxembourg | ||
11:00 20mTalk | Mutation testing of PL/SQL programs Journal-First Papers | ||
11:20 20mTalk | How Closely are Common Mutation Operators Coupled to Real Faults? Research Papers | ||
11:40 20mTalk | Batching Non-Conflicting Mutations for Efficient, Safe, Parallel Mutation Analysis in Rust Research Papers | ||
12:00 20mTalk | MUTAGEN: Reliable Coverage-Guided, Property-Based Testing using Exhaustive Mutations Research Papers Agustín Mista Chalmers University of Technology, Alejandro Russo Chalmers University of Technology, Sweden |