Dispatch chart
Subscriber scaling
Compare one, five, and twenty typed handlers with wildcard-plus-typed dispatch.
Open dispatch timings →Execution latency
Read the cost in time.
Lower is better. Each value is the median time for one precisely named operation; setup, validation, calibration, warmup, verification, reporting, and file I/O are outside the plotted timing boundary.
—typed emit · 1 handler
—typed emit · 5 handlers
—wildcard + 5 typed
—once registration + emit cycle
Dispatch scaling
Median execution time per emit · lower is better
Loading current CI benchmark…
Loading timing evidence…
Lifecycle charts
Mutation and state
See once-plus-emit, on-plus-off, list snapshot, and register-five-plus-reset cycles without mixing unlike units.
Open lifecycle timings →Timing boundary
What counts
Inspect the fixed-iteration method, exclusions, environment, raw samples, and exact code timed.
Open methodology →Run the same harness
npm run benchmark
npm run benchmark:smokeMachine-readable evidence
Schema v2 publishes elapsed nanoseconds, operation counts, per-operation latency, median, p25, p75, min, max, exact workload boundaries, and machine context.
Open benchmark JSON →These are measurements, not promises.
Compare commits on the same runtime and hardware. The charts make scaling and regressions readable; application handlers and real workloads will add their own execution time.