mirror of
https://github.com/windmill-labs/windmill.git
synced 2026-10-17 16:03:06 +00:00
The total duration of a for-loop/branchall group was computed as the naive sum of all iteration durations. This is wrong for parallel execution and doesn't account for orchestration overhead. Instead, compute actual wall-clock time as max(completed_at) - min(started_at). Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>