comphead commented on code in PR #15418:
URL: https://github.com/apache/datafusion/pull/15418#discussion_r2012314429


##########
datafusion/physical-plan/src/joins/hash_join.rs:
##########
@@ -791,46 +791,52 @@ impl ExecutionPlan for HashJoinExec {
             );
         }
 
+        if self.mode == PartitionMode::Auto {
+            return plan_err!(
+                "Invalid HashJoinExec, unsupported PartitionMode {:?} in 
execute()",
+                PartitionMode::Auto
+            );
+        }
+
         let join_metrics = BuildProbeJoinMetrics::new(partition, 
&self.metrics);
         let left_fut = match self.mode {
-            PartitionMode::CollectLeft => self.left_fut.once(|| {
-                let reservation =
-                    
MemoryConsumer::new("HashJoinInput").register(context.memory_pool());
-                collect_left_input(
-                    None,
-                    self.random_state.clone(),
-                    Arc::clone(&self.left),
-                    on_left.clone(),
-                    Arc::clone(&context),
-                    join_metrics.clone(),
-                    reservation,
-                    need_produce_result_in_final(self.join_type),
-                    self.right().output_partitioning().partition_count(),
-                )
-            }),
+            PartitionMode::CollectLeft => {
+                let left = 
coalesce_partitions_if_needed(Arc::clone(&self.left));
+                let left_stream = left.execute(0, Arc::clone(&context))?;
+
+                self.left_fut.once(|| {
+                    let reservation = MemoryConsumer::new("HashJoinInput")
+                        .register(context.memory_pool());
+
+                    collect_left_input(
+                        self.random_state.clone(),
+                        left_stream,
+                        on_left.clone(),
+                        join_metrics.clone(),
+                        reservation,
+                        need_produce_result_in_final(self.join_type),
+                        self.right().output_partitioning().partition_count(),
+                    )
+                })
+            }
             PartitionMode::Partitioned => {
+                let left_stream = self.left.execute(partition, 
Arc::clone(&context))?;
+
                 let reservation =
                     MemoryConsumer::new(format!("HashJoinInput[{partition}]"))
                         .register(context.memory_pool());
 
                 OnceFut::new(collect_left_input(
-                    Some(partition),
                     self.random_state.clone(),
-                    Arc::clone(&self.left),
+                    left_stream,
                     on_left.clone(),
-                    Arc::clone(&context),
                     join_metrics.clone(),
                     reservation,
                     need_produce_result_in_final(self.join_type),
                     1,
                 ))
             }
-            PartitionMode::Auto => {
-                return plan_err!(
-                    "Invalid HashJoinExec, unsupported PartitionMode {:?} in 
execute()",
-                    PartitionMode::Auto
-                );
-            }
+            PartitionMode::Auto => unreachable!(),

Review Comment:
   Imho it is better to keep a result instead of a crash. If this functionality 
is regressed so the users gonna face the  app crash



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: github-unsubscr...@datafusion.apache.org

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscr...@datafusion.apache.org
For additional commands, e-mail: github-h...@datafusion.apache.org

Reply via email to