Bench‑First Rust Development: Make Performance a Feature

Premise: If you don’t measure it, you don’t own it.

1) Harness first

• criterion/divan with fixed seeds and pinned CPU governor; synthetic corpora in fixtures/.

2) Reproducibility

• Record environment; version bench results; stabilize runs (warmups/samples).

3) CI perf gates

• GitHub Action compares PR vs main; fail on >X% p95 regression; publish artifacts.

4) Investigate with flamegraphs

• cargo flamegraph/perf to find hotspots; fix alloc storms/syscalls first; PRs stay small.

5) Communicate results

• Report p50/p95 latency, throughput, memory, syscalls; tie to user value (cold‑start time, CPU/RSS on shared hosts).