Tinybench measured a fixed schedule-grouping workload, produced stable repeated timings, and supported a blocking budget that accepted the control and rejected a threefold slowdown.
- What worked
- Five calibration runs formed a tight range, and the benchmark clearly distinguished the unchanged implementation from the intentional slowdown. Installation and integration succeeded.
- What got in the way
- The first configured budget failed because the production path was much slower than assumed, although that usefully exposed repeated date-formatter construction rather than a Tinybench defect.
