Apache DataFusion SQL Query Engine
perf: skip re-slicing window partition batches with nothing to prune (#24047)
Neil Conway47ca49043aeb
diff --git a/datafusion/physical-plan/src/windows/bounded_window_agg_exec.rs b/datafusion/physical-plan/src/windows/bounded_window_agg_exec.rsindex 07751a70ec..680d5b657e 100644--- a/datafusion/physical-plan/src/windows/bounded_window_agg_exec.rs+++ b/datafusion/physical-plan/src/windows/bounded_window_agg_exec.rs@@ -1188,10 +1188,15 @@ impl BoundedWindowAggStream {// Retract no longer needed parts during window calculations from partition batch:for (partition_row, n_prune) in n_prune_each_partition.iter() {let pb_state = &mut self.partition_buffers[partition_row];+ pb_state.n_out_row = 0;++ // If there is nothing to prune, leave the batch as-is+ if *n_prune == 0 {+ continue;+ }let batch = &pb_state.record_batch;pb_state.record_batch = batch.slice(*n_prune, batch.num_rows() - n_prune);- pb_state.n_out_row = 0;// Update state indices since we have pruned some rows from the beginning:for window_agg_state in self.window_agg_states.iter_mut() {