Exploratory, idiom-vs-idiom micro-benchmarks from the
org.apache.groovy.adhoc package — kept out of the core
regression suites on purpose. Each chart shows the latest
run only (these compare idioms against each other, not Groovy over
time). Throughput is ops/µs (higher is better);
allocation is gc.alloc.rate.norm bytes/op
(lower is better). Both use a log y-axis — per family the
workload sizes span roughly two decades, so a linear axis would flatten the
larger sizes into indistinguishable slivers; for allocation, bars at the
0 floor mean effectively zero allocation. Bars are grouped by
workload size. Use the
toggle to switch between indy and classic call
sites. Per-run history (with regression alerting) lives under the
per-bucket dashboards linked below.