Skip to content

fix: preserve byte-size statistics in AggregateExec #18885

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Nov 25, 2025
+182 −62
Merged

fix: preserve byte-size statistics in AggregateExec #18885

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
5916ebb
fix: preserve byte-size statistics in AggregateExec
Tamar-Posen Nov 22, 2025
8f7fad7
Merge branch 'main' into fix-aggregate-bytes-stats
Dandandan Nov 24, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
102 changes: 54 additions & 48 deletions datafusion/core/tests/dataframe/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3302,30 +3302,33 @@ async fn test_count_wildcard_on_where_scalar_subquery() -> Result<()> {
assert_snapshot!(
pretty_format_batches(&sql_results).unwrap(),
@r"
+---------------+-------------------------------------------------------------------------------------------------------------------------+
Copy link
Contributor Author

@Tamar-Posen Tamar-Posen Nov 23, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The physical plan changed bc join selection now uses available byte stats.
Which leads to a more optimal join order - ensuring the smaller side is chosen as the build side.

| plan_type | plan |
+---------------+-------------------------------------------------------------------------------------------------------------------------+
| logical_plan | Projection: t1.a, t1.b |
| | Filter: CASE WHEN __scalar_sq_1.__always_true IS NULL THEN Int64(0) ELSE __scalar_sq_1.count(*) END > Int64(0) |
| | Projection: t1.a, t1.b, __scalar_sq_1.count(*), __scalar_sq_1.__always_true |
| | Left Join: t1.a = __scalar_sq_1.a |
| | TableScan: t1 projection=[a, b] |
| | SubqueryAlias: __scalar_sq_1 |
| | Projection: count(Int64(1)) AS count(*), t2.a, Boolean(true) AS __always_true |
| | Aggregate: groupBy=[[t2.a]], aggr=[[count(Int64(1))]] |
| | TableScan: t2 projection=[a] |
| physical_plan | FilterExec: CASE WHEN __always_true@3 IS NULL THEN 0 ELSE count(*)@2 END > 0, projection=[a@0, b@1] |
| | CoalesceBatchesExec: target_batch_size=8192 |
| | HashJoinExec: mode=CollectLeft, join_type=Left, on=[(a@0, a@1)], projection=[a@0, b@1, count(*)@2, __always_true@4] |
| | DataSourceExec: partitions=1, partition_sizes=[1] |
| | ProjectionExec: expr=[count(Int64(1))@1 as count(*), a@0 as a, true as __always_true] |
| | AggregateExec: mode=FinalPartitioned, gby=[a@0 as a], aggr=[count(Int64(1))] |
| | CoalesceBatchesExec: target_batch_size=8192 |
| | RepartitionExec: partitioning=Hash([a@0], 4), input_partitions=1 |
| | AggregateExec: mode=Partial, gby=[a@0 as a], aggr=[count(Int64(1))] |
| | DataSourceExec: partitions=1, partition_sizes=[1] |
| | |
+---------------+-------------------------------------------------------------------------------------------------------------------------+
+---------------+------------------------------------------------------------------------------------------------------------------------------+
| plan_type | plan |
+---------------+------------------------------------------------------------------------------------------------------------------------------+
| logical_plan | Projection: t1.a, t1.b |
| | Filter: CASE WHEN __scalar_sq_1.__always_true IS NULL THEN Int64(0) ELSE __scalar_sq_1.count(*) END > Int64(0) |
| | Projection: t1.a, t1.b, __scalar_sq_1.count(*), __scalar_sq_1.__always_true |
| | Left Join: t1.a = __scalar_sq_1.a |
| | TableScan: t1 projection=[a, b] |
| | SubqueryAlias: __scalar_sq_1 |
| | Projection: count(Int64(1)) AS count(*), t2.a, Boolean(true) AS __always_true |
| | Aggregate: groupBy=[[t2.a]], aggr=[[count(Int64(1))]] |
| | TableScan: t2 projection=[a] |
| physical_plan | FilterExec: CASE WHEN __always_true@3 IS NULL THEN 0 ELSE count(*)@2 END > 0, projection=[a@0, b@1] |
| | RepartitionExec: partitioning=RoundRobinBatch(4), input_partitions=1 |
| | ProjectionExec: expr=[a@2 as a, b@3 as b, count(*)@0 as count(*), __always_true@1 as __always_true] |
| | CoalesceBatchesExec: target_batch_size=8192 |
| | HashJoinExec: mode=CollectLeft, join_type=Right, on=[(a@1, a@0)], projection=[count(*)@0, __always_true@2, a@3, b@4] |
| | CoalescePartitionsExec |
| | ProjectionExec: expr=[count(Int64(1))@1 as count(*), a@0 as a, true as __always_true] |
| | AggregateExec: mode=FinalPartitioned, gby=[a@0 as a], aggr=[count(Int64(1))] |
| | CoalesceBatchesExec: target_batch_size=8192 |
| | RepartitionExec: partitioning=Hash([a@0], 4), input_partitions=1 |
| | AggregateExec: mode=Partial, gby=[a@0 as a], aggr=[count(Int64(1))] |
| | DataSourceExec: partitions=1, partition_sizes=[1] |
| | DataSourceExec: partitions=1, partition_sizes=[1] |
| | |
+---------------+------------------------------------------------------------------------------------------------------------------------------+
"
);

Expand Down Expand Up @@ -3357,30 +3360,33 @@ async fn test_count_wildcard_on_where_scalar_subquery() -> Result<()> {
assert_snapshot!(
pretty_format_batches(&df_results).unwrap(),
@r"
+---------------+-------------------------------------------------------------------------------------------------------------------------+
| plan_type | plan |
+---------------+-------------------------------------------------------------------------------------------------------------------------+
| logical_plan | Projection: t1.a, t1.b |
| | Filter: CASE WHEN __scalar_sq_1.__always_true IS NULL THEN Int64(0) ELSE __scalar_sq_1.count(*) END > Int64(0) |
| | Projection: t1.a, t1.b, __scalar_sq_1.count(*), __scalar_sq_1.__always_true |
| | Left Join: t1.a = __scalar_sq_1.a |
| | TableScan: t1 projection=[a, b] |
| | SubqueryAlias: __scalar_sq_1 |
| | Projection: count(*), t2.a, Boolean(true) AS __always_true |
| | Aggregate: groupBy=[[t2.a]], aggr=[[count(Int64(1)) AS count(*)]] |
| | TableScan: t2 projection=[a] |
| physical_plan | FilterExec: CASE WHEN __always_true@3 IS NULL THEN 0 ELSE count(*)@2 END > 0, projection=[a@0, b@1] |
| | CoalesceBatchesExec: target_batch_size=8192 |
| | HashJoinExec: mode=CollectLeft, join_type=Left, on=[(a@0, a@1)], projection=[a@0, b@1, count(*)@2, __always_true@4] |
| | DataSourceExec: partitions=1, partition_sizes=[1] |
| | ProjectionExec: expr=[count(*)@1 as count(*), a@0 as a, true as __always_true] |
| | AggregateExec: mode=FinalPartitioned, gby=[a@0 as a], aggr=[count(*)] |
| | CoalesceBatchesExec: target_batch_size=8192 |
| | RepartitionExec: partitioning=Hash([a@0], 4), input_partitions=1 |
| | AggregateExec: mode=Partial, gby=[a@0 as a], aggr=[count(*)] |
| | DataSourceExec: partitions=1, partition_sizes=[1] |
| | |
+---------------+-------------------------------------------------------------------------------------------------------------------------+
+---------------+------------------------------------------------------------------------------------------------------------------------------+
| plan_type | plan |
+---------------+------------------------------------------------------------------------------------------------------------------------------+
| logical_plan | Projection: t1.a, t1.b |
| | Filter: CASE WHEN __scalar_sq_1.__always_true IS NULL THEN Int64(0) ELSE __scalar_sq_1.count(*) END > Int64(0) |
| | Projection: t1.a, t1.b, __scalar_sq_1.count(*), __scalar_sq_1.__always_true |
| | Left Join: t1.a = __scalar_sq_1.a |
| | TableScan: t1 projection=[a, b] |
| | SubqueryAlias: __scalar_sq_1 |
| | Projection: count(*), t2.a, Boolean(true) AS __always_true |
| | Aggregate: groupBy=[[t2.a]], aggr=[[count(Int64(1)) AS count(*)]] |
| | TableScan: t2 projection=[a] |
| physical_plan | FilterExec: CASE WHEN __always_true@3 IS NULL THEN 0 ELSE count(*)@2 END > 0, projection=[a@0, b@1] |
| | RepartitionExec: partitioning=RoundRobinBatch(4), input_partitions=1 |
| | ProjectionExec: expr=[a@2 as a, b@3 as b, count(*)@0 as count(*), __always_true@1 as __always_true] |
| | CoalesceBatchesExec: target_batch_size=8192 |
| | HashJoinExec: mode=CollectLeft, join_type=Right, on=[(a@1, a@0)], projection=[count(*)@0, __always_true@2, a@3, b@4] |
| | CoalescePartitionsExec |
| | ProjectionExec: expr=[count(*)@1 as count(*), a@0 as a, true as __always_true] |
| | AggregateExec: mode=FinalPartitioned, gby=[a@0 as a], aggr=[count(*)] |
| | CoalesceBatchesExec: target_batch_size=8192 |
| | RepartitionExec: partitioning=Hash([a@0], 4), input_partitions=1 |
| | AggregateExec: mode=Partial, gby=[a@0 as a], aggr=[count(*)] |
| | DataSourceExec: partitions=1, partition_sizes=[1] |
| | DataSourceExec: partitions=1, partition_sizes=[1] |
| | |
+---------------+------------------------------------------------------------------------------------------------------------------------------+
"
);

Expand Down
4 changes: 2 additions & 2 deletions datafusion/core/tests/physical_optimizer/partition_statistics.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -627,7 +627,7 @@ mod test {

let expected_p0_statistics = Statistics {
num_rows: Precision::Inexact(2),
total_byte_size: Precision::Absent,
total_byte_size: Precision::Inexact(110),
column_statistics: vec![
ColumnStatistics {
null_count: Precision::Absent,
Expand All @@ -645,7 +645,7 @@ mod test {

let expected_p1_statistics = Statistics {
num_rows: Precision::Inexact(2),
total_byte_size: Precision::Absent,
total_byte_size: Precision::Inexact(110),
column_statistics: vec![
ColumnStatistics {
null_count: Precision::Absent,
Expand Down
Loading