v1.0.0 - Initial Release

Embedded Integration Benchmark Toolkit v1.0.0

Empirical benchmarks for embedded Linux IPC architecture decisions — measure before you commit to a pattern.

Benchmarks

• SQLite WAL contention — C writer + Go reader, 6 scenarios (WAL vs rollback, 33–330 w/s, 1–3 concurrent readers), 10-minute sustained test
• inotify sentinel files — Go writer + C watcher, atomic rename delivery, dispatch/processing/pipeline latency
• Unix domain sockets — Go client + C server, identical payloads for direct comparison with inotify
• Shared memory (mmap + FIFO) — Go writer + C reader, ARMv7 memory barriers, sequence integrity tracking
• inotify reliability — 5 failure-mode scenarios: queue overflow, I/O storm, rapid overwrite, fsync race, tmpfs vs block device

Tooling

• bench CLI — YAML-driven orchestration, schema introspection, runtime config bridge
• Verdict-first report generator with pass/fail scorecard and threshold gating
• GitHub Action — uses: mgmacri/embedded-integration-benchmark-toolkit@v1
• Docker cross-compilation (ARM Linux armv7l via arm-linux-gnueabihf-gcc)
• Deploy/collect scripts for target device workflow

GitHub Action Usage

- uses: mgmacri/embedded-integration-benchmark-toolkit@v1
  with:
    results: path/to/results/
    config: bench.yaml
    gate: 'true'

Outputs: verdict (pass/fail), report-path, transports, transport-count

See CI-INTEGRATION.md for full documentation.