💻 AI in the role of a "trial developer" for Java
Peter Lawrey used Codex/GPT-5.5 to create the JLBH benchmark for the Chronicle-FIX library. This allowed for evaluating the impact of garbage collectors on message round-trip latency, demonstrating a throughput of 50k messages per second.
🌍 Using AI allows for identifying flaws in documentation and APIs due to the absence of an experienced developer's biases. This opens pathways for automated testing of closed-source systems.
👤 AI can effectively write specific low-latency benchmarks by working solely based on documentation and code examples.
Source 1: https://blog.vanillajava.blog/2026/06/testing-java-memory-management-with.html Source 2: https://www.linkedin.com/posts/peterlawrey_testing-java-memory-management-with-chronicle-fix-activity-7468592211799031808-hi7-