test262-pass-rates

Test262 Pass Rates (js-host vs standalone)

Prints the current committed test262 conformance numbers for both compile targets side by side. Reads two committed, CI-refreshed summary files directly — no test run, no network fetch, near-instant.

Data sources

• js-host (gc target): benchmarks/results/test262-current.json — refreshed by the promote-baseline job in test262-sharded.yml on every push to main. Use .summary (or .official_summary if present) for pass/total, and .summary.host_free_pass for the narrower "passes with zero host imports" count (not the same as a standalone-target run — see below).
• standalone target: benchmarks/results/test262-standalone-highwater.json — refreshed alongside the js-host baseline (same sha/generated_at). Use .official_pass / .official_total for the genuine standalone-target compile+run pass rate.

Both files are committed to the repo (small, ~KB-scale) and always reflect the last successful push-to-main run — check .timestamp/.generated_at and .sha against git -C /workspace log -1 --format=%H to confirm freshness; if the sha doesn't match current main, note that the numbers are as-of the last CI run, not live.

Steps

1. Read both files:

node -e "
const host = require('/workspace/benchmarks/results/test262-current.json');
const sa = require('/workspace/benchmarks/results/test262-standalone-highwater.json');
const h = host.official_summary ?? host.summary;
const hostPct = (100 * h.pass / h.total).toFixed(1);
const saPct = (100 * sa.official_pass / sa.official_total).toFixed(1);
const gap = h.pass - sa.official_pass;
console.log('js-host (gc):   ' + h.pass + ' / ' + h.total + '  (' + hostPct + '%)');
console.log('standalone:     ' + sa.official_pass + ' / ' + sa.official_total + '  (' + saPct + '%)');
console.log('honest gap:     ' + gap + ' tests pass under js-host but not standalone');
console.log('baseline sha:   ' + host.baseline_sha);
console.log('generated at:   ' + host.timestamp);
"