Beyond Throughput — v1.0.1

Corrective reproducible release of Beyond Throughput: A Reproducible Benchmark of Java Caches in a Tomcat Lifecycle.

This release fixes cross-platform checksum validation and adds the community, CI, and security scaffolding introduced in PR #4. The benchmark protocol (v4.2), canonical campaign, numerical results, and scientific conclusions are unchanged.

Reproduce

git clone --branch v1.0.1 https://github.com/tbuffagni/brainy-java-cache-lab.git
cd brainy-java-cache-lab
docker compose up --build

The canonical campaign completed 36 independent JVM runs, 180 lifecycle cycles, 360 timed windows, and 18,934 semantic checks. The separate lifecycle analysis reproduces the worker-retention pattern with JCS 3.2.1 and does not reproduce it with the examined JCS 4 snapshot.

Included assets: Italian and English papers, the results workbook, the compact evidence archive, and the complete press package.

Beyond Throughput — v1.0.0

First reproducible release of Beyond Throughput: A Reproducible Benchmark of Java Caches in a Tomcat Lifecycle.

This release freezes the paper, experimental protocol, benchmark implementation, result workbook, and complete evidence archive used by the study.

Included assets

• English paper (PDF)
• Italian paper (PDF)
• Excel workbook with tables and charts
• Complete evidence archive with raw JSON, derived analysis, CSV observations, build log, diagnostics, and validation checkpoints
• SHA-256 manifest for all uploaded assets

Reproduction

Clone the exact release together with the pinned Apache Commons JCS source revision:

git clone --recurse-submodules --branch v1.0.0 https://github.com/tbuffagni/brainy-java-cache-lab.git

The canonical campaign completed 36 independent JVM processes, 180 Tomcat lifecycle cycles, and 360 timed workload windows. The automated validator completed 18,934 checks with no errors or warnings.

Throughput, latency, semantic correctness, and lifecycle behavior are separate results. The measurements describe the configured application path and environment; they are not presented as a universal ranking of cache providers.

The worker-thread retention pattern reproduced with JCS 3.2.1 was not reproduced with the pinned JCS 4 revision. This conclusion is limited to the specific pattern and protocol tested.