Uh oh!
There was an error while loading. Please reload this page.
feat: change Expr OuterReferenceColumn and Alias to Box type for reducing expr struct size - #16771
Closed
zhuqi-lucas wants to merge 2 commits into
Closed
feat: change Expr OuterReferenceColumn and Alias to Box type for reducing expr struct size#16771zhuqi-lucas wants to merge 2 commits into
zhuqi-lucas wants to merge 2 commits into
Conversation
alamb
reviewed
Jul 14, 2025
alamb
left a comment
Contributor
There was a problem hiding this comment.
Thanks @zhuqi-lucas -- this looks quite nice
I think it is an API change so I will mark the PR as such and I think it is a good improvement
However, given it is an API change I think we should
- Wait until DataFusion 50.0.0 (let's not make any more changes in DataFusion 49 that we are starting to prepare for)
- Update the upgrading guide in https://datafusion.apache.org/library-user-guide/upgrading.html to explain to people hw to change things
| 10)--------------Projection: generate_series(Int64(1), CAST(outer_ref(t1.t1_int) AS Int64)) AS __unnest_placeholder(generate_series(Int64(1),outer_ref(t1.t1_int))) | ||
| 11)----------------EmptyRelation | ||
| physical_plan_error This feature is not implemented: Physical plan does not support logical expression OuterReferenceColumn(UInt32, Column { relation: Some(Bare { table: "t1" }), name: "t1_int" }) | ||
| physical_plan_error This feature is not implemented: Physical plan does not support logical expression OuterReferenceColumn((UInt32, Column { relation: Some(Bare { table: "t1" }), name: "t1_int" })) |
Contributor
There was a problem hiding this comment.
do you know why the plan changes? I don't think the extra () really adds much -- can we restore the original?
ContributorAuthor
There was a problem hiding this comment.
I changed it to OuterReference in latest PR, so it will use OuterReference instead of this change now, thanks!
ContributorAuthor
There was a problem hiding this comment.
Next step, we may can add debug/fmt for OuterReference to restore the original behaviour.
ContributorAuthor
There was a problem hiding this comment.
Restore all slt in latest PR, no diff from main branch.
| // `Box`ing the fields to make `Expr` smaller | ||
| // See https://github.com/apache/datafusion/issues/16199 for details | ||
| assert_eq!(size_of::<Expr>(), 128); | ||
| assert_eq!(size_of::<Expr>(), 112); |
Contributor
There was a problem hiding this comment.
This change saves 16 bytes per Expr. Nice.
I will run some planning benchmarks and see if we can see any difference
ContributorAuthor
There was a problem hiding this comment.
Thank you @alamb , i am doing more amazing experiment, try to reduce from 128 to 80, so we can save 48 bytes per Expr!
| pub enum Expr { | ||
| /// An expression with a specific name. | ||
| Alias(Alias), | ||
| Alias(Box<Alias>), |
Contributor
There was a problem hiding this comment.
Another change that might be a bit less impactful would be to box the fields of Alias instead
So Alias {
expr: Box
..
}
It may be just as bad / worse though
| /// A placeholder which holds a reference to a qualified field | ||
| /// in the outer query, used for correlated sub queries. | ||
| OuterReferenceColumn(DataType, Column), | ||
| OuterReferenceColumn(Box<(DataType, Column)>), |
Contributor
There was a problem hiding this comment.
If we are going to make this change anyways, can we also pull this into a named struct rather than a unnamed tuple
like
structOuterReference{// fields here}enumExpr{
... OuterReferenceColumn(OuterReference),
...
}ContributorAuthor
There was a problem hiding this comment.
Thank you @alamb for this good suggestion, addressed in latest PR.
alamb
commented
Jul 14, 2025
Contributor
🤖 |
alamb
commented
Jul 14, 2025
Contributor
🤖: Benchmark completed Details |
zhuqi-lucas
commented
Jul 15, 2025
ContributorAuthor
Thank you @alamb for review and i agree that we can do this for Datafusion 50.0.0, and since we have enough time, i want to do more improvement, try to reduce Expr more, from size 128 to 80, i am in progress now. |
zhuqi-lucas
commented
Jul 15, 2025
ContributorAuthor
Updated: Successfully changed the size from 128 to 80 in latest PR. |
alamb
commented
Jul 15, 2025
Contributor
🤖 |
alamb
commented
Jul 15, 2025
Contributor
🤖: Benchmark completed Details |
alamb
commented
Jul 15, 2025
Contributor
🤖 |
alamb
commented
Jul 15, 2025
Contributor
🤖: Benchmark completed Details |
zhuqi-lucas
commented
Jul 16, 2025
ContributorAuthor
Thank you @alamb for benchmark, interesting, it seems the performance decrease from 112 size to 80, i will investigate it, if i can't find the root cause, i will revert to 112 bytes first! |
zhuqi-lucas
commented
Jul 16, 2025
ContributorAuthor
Strange, i can't reproduce this in my local: with_param_values_many_columns 1.00129.8±0.74µs ? ?/sec 1.06137.2±0.79µs |
kosiew
reviewed
Jul 16, 2025
kosiew
left a comment
Contributor
There was a problem hiding this comment.
👍
Just some nit for your review.
Comment on lines
+260
to
+262
| Expr::Alias(alias_box) => { | ||
| // alias_box: Box<Alias> | ||
| let Alias { expr, name, .. } = *alias_box; |
Contributor
There was a problem hiding this comment.
nit - use boxed_alias instead, for consistency
ContributorAuthor
There was a problem hiding this comment.
Thank you @kosiew , addressed in latest PR.
Comment on lines
+140
to
+142
| if let Expr::Alias(alias_box) = expr { | ||
| // alias_box: &Box<Alias>, so alias_box.as_ref() is &Alias | ||
| let alias: &Alias = alias_box.as_ref(); |
Comment on lines
+249
to
+250
| Expr::Alias(alias_box) => { | ||
| let Alias { expr, .. } = alias_box.as_ref(); |
| stack.push(*left); | ||
| } | ||
| Expr::Alias(Alias { expr, .. }) => stack.push(*expr), | ||
| Expr::Alias(alias_box) => stack.push(*alias_box.expr), |
zhuqi-lucas
commented
Jul 16, 2025
ContributorAuthor
Thank you @kosiew for review and suggestion, addressed in latest PR. |
xudong963
commented
Jul 16, 2025
Member
After opening the DF50.0.0 release issue, you can add it to the list |
alamb
commented
Jul 16, 2025
Contributor
I will rerun |
alamb
commented
Jul 16, 2025
Contributor
🤖 |
zhuqi-lucas pushed a commit
to zhuqi-lucas/arrow-datafusion
that referenced
this pull request
Feb 10, 2026
Box three large variants in the Expr enum to reduce its memory footprint: - Alias(Alias) -> Alias(Box<Alias>) - Column(Column) -> Column(Box<Column>) - OuterReferenceColumn(FieldRef, Column) -> OuterReferenceColumn(Box<OuterReference>) This reduces the Expr enum size from 112 bytes to 80 bytes (28% reduction), which improves cache locality and reduces memory usage across the query engine since Expr is one of the most frequently used types. Introduces a new OuterReference named struct to replace the tuple variant, improving readability and maintainability. Based on apache#16771, rebased and updated against the latest master. https://claude.ai/code/session_018FV5mepXRLnBVzzWv8ihgb
zhuqi-lucasforce-pushed
the
reduce_expr_size
branch
from
February 11, 2026 07:44
569079b to
b5716d8CompareContributor
There was a problem hiding this comment.
Pull request overview
This PR reduces the size of the Expr enum from 128 bytes to 80 bytes by boxing the Alias and OuterReferenceColumn variants. This is a continuation of ongoing efforts to optimize memory usage in DataFusion's expression system.
Changes:
- Changed
Expr::Alias(Alias)toExpr::Alias(Box<Alias>) - Changed
Expr::OuterReferenceColumn(FieldRef, Column)toExpr::OuterReferenceColumn(Box<OuterReference>)whereOuterReferenceis a new struct containing the field and column - Updated all pattern matching and construction sites throughout the codebase to work with the boxed types
- Updated test expectations for error messages that display
OuterReferenceColumn - Updated the size test to verify the new 80-byte size
Reviewed changes
Copilot reviewed 58 out of 58 changed files in this pull request and generated no comments.
Show a summary per file
| File | Description |
|---|---|
| datafusion/expr/src/expr.rs | Core type changes: boxed Alias and OuterReferenceColumn, added OuterReference struct, updated size test |
| datafusion/expr/src/lib.rs | Exported OuterReference type |
| datafusion/expr/src/*.rs | Updated pattern matching and expression construction for boxed types |
| datafusion/optimizer/src/*.rs | Updated optimizer rules to work with boxed types |
| datafusion/sql/src/*.rs | Updated SQL planning to construct boxed expressions |
| datafusion/substrait/src/*.rs | Updated Substrait conversion to handle boxed types |
| datafusion/proto/src/*.rs | Updated protobuf serialization/deserialization |
| datafusion/physical-expr/src/*.rs | Updated physical expression planning |
| datafusion/core/src/*.rs | Updated core dataframe and execution code |
| datafusion/functions*/src/*.rs | Updated function implementations |
| datafusion/**/tests/*.rs | Updated test code and test expectations |
| datafusion/sqllogictest/test_files/*.slt | Updated expected error messages for OuterReferenceColumn display format |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
zhuqi-lucas
commented
Feb 11, 2026
ContributorAuthor
run sql_planner |
zhuqi-lucas
commented
Feb 11, 2026
ContributorAuthor
run benchmark sql_planner |
alamb-ghbot
commented
Feb 11, 2026
🤖 |
zhuqi-lucas
commented
Feb 11, 2026
ContributorAuthor
I revive this PR by claude help. cc @alamb And let's see if benchmark will show improvement. |
alamb-ghbot
commented
Feb 11, 2026
🤖: Benchmark completed Details |
zhuqi-lucas
commented
Feb 11, 2026
ContributorAuthor
Seems still no improvement in latest benchmark. |
zhuqi-lucasforce-pushed
the
reduce_expr_size
branch
from
February 11, 2026 14:07
edf1eaf to
3941abaComparezhuqi-lucas
commented
Feb 11, 2026
ContributorAuthor
run benchmark sql_planner |
zhuqi-lucas
commented
Feb 11, 2026
ContributorAuthor
run benchmark sql_planner |
alamb
commented
Feb 11, 2026
Contributor
show benchmark queue |
alamb-ghbot
commented
Feb 11, 2026
🤖 Hi @alamb, you asked to view the benchmark queue (#16771 (comment)).
|
zhuqi-lucas
commented
Feb 11, 2026
ContributorAuthor
Thanks @alamb, i see, there are already 2 in queue.
|
Contributor
👍 -- I just wanted to double check that the runner hadn't gotten stuck again (it sometimes does break / get restarted and needs manual intervention) |
alamb-ghbot
commented
Feb 11, 2026
🤖 |
alamb-ghbot
commented
Feb 11, 2026
🤖: Benchmark completed Details |
alamb-ghbot
commented
Feb 11, 2026
🤖 |
alamb-ghbot
commented
Feb 11, 2026
🤖: Benchmark completed Details |
alamb
commented
Mar 13, 2026
Contributor
I think my conclusion is that this PR doesn't really decrease planning time measurably, so marking as a draft for now |
alamb
marked this pull request as draft
March 13, 2026 18:29
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Which issue does this PR close?
Continue to reduce the Expr struct size.
Rationale for this change
Continue to reduce the Expr struct size.
What changes are included in this PR?
Continue to reduce the Expr struct size.
Are these changes tested?
Yes
The size reduce from 128 to 112
Updated:
Continue reduce to 80 now!
Are there any user-facing changes?
No