Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Watch a 15-minute conference presentation from OOPSLA 2025 that introduces two novel approaches to significantly improve the efficiency of bytecode frontends in Java static analysis frameworks. Learn about the critical role of bytecode frontends in transforming complex stack-based Java bytecode into analyzable register-based, typed 3-address code representation, and discover how current dominant frameworks like Soot and WALA face efficiency limitations when processing large-scale Java applications. Explore the researchers' identification of common patterns in bytecode that enable more efficient processing and their findings about redundant computations in traditional type inference algorithms. Understand the two innovative solutions presented: pattern-aware 3-address code translation and pruning-based type inference, which together form a new frontend implementation in the Tai-e static analysis framework. Examine experimental results demonstrating performance improvements of 14.2×, 14.5×, and 75.2× faster processing compared to Soot, WALA, and SootUp respectively, along with superior reliability in bytecode processing. Gain insights into how this new approach can generate SSA IR to enhance usability for various static analysis techniques, providing a more robust foundation for Java static analysis with comprehensive artifact availability and reproducible results.
