MCPcopy Create free account
hub / github.com/idank/explainshell / compare_runs

Function compare_runs

tests/evals/render/render_eval.py:441–455  ·  view source on GitHub ↗
(args: argparse.Namespace)

Source from the content-addressed store, hash-verified

439
440
441def compare_runs(args: argparse.Namespace) -> int:
442 base_dir = Path(args.baseline)
443 current_dir = Path(args.current)
444 baseline, current, paths, suspicious = _comparison_data(base_dir, current_dir)
445 report = _summarize_comparison(
446 base_dir, current_dir, baseline, current, suspicious, paths
447 )
448 report_path = current_dir / "comparison.md"
449 report_path.write_text(report)
450 print(report)
451 print(f"comparison report: {report_path}")
452
453 if suspicious and args.fail_on_suspicious:
454 return 1
455 return 0
456
457
458def _artifact_path(run_dir: Path, page: dict[str, Any], kind: str, suffix: str) -> Path:

Callers

nothing calls this directly

Calls 2

_comparison_dataFunction · 0.85
_summarize_comparisonFunction · 0.85

Tested by

no test coverage detected