Avoid creating new RecordBatches to simplify expressions#20534
Merged
Conversation
alamb
commented
Feb 24, 2026
| let col = new_null_array(&DataType::Null, 1); | ||
| Ok(RecordBatch::try_new(dummy_schema, vec![col])?) | ||
| pub(crate) fn create_dummy_batch() -> Result<&'static RecordBatch> { | ||
| static DUMMY_BATCH: std::sync::OnceLock<Result<RecordBatch>> = |
Contributor
Author
There was a problem hiding this comment.
The idea is to compute the result once and then reuse it each time rather than create a new batch and array each time
Contributor
Author
|
I don't really expect to be able to measure the improvement here but I'll run benchmarks to check |
Contributor
Author
|
run benchmark sql_planner |
AdamGS
approved these changes
Feb 24, 2026
AdamGS
left a comment
AdamGS
left a comment
Contributor
There was a problem hiding this comment.
LGTM :) I think I had this at some point, but I couldn't measure a difference, even though it's obviously strictly less work
|
🤖 |
|
🤖: Benchmark completed Details
|
Contributor
Author
Yeah, I agree I will be surprised if this shows up in the traces. The first benchmark run seems to show improvements but I suspect it is just noise. I will rerun to see if I can reproduce |
Contributor
Author
|
run benchmark sql_planner |
|
🤖 |
|
🤖: Benchmark completed Details
|
Contributor
Author
|
TLDR is this doesn't show up in any specific benchmarksing |
Contributor
|
@alamb shall we close this then? |
Contributor
Author
I personally think it is a worthwhile opimtization as it doesn't add much complexity and saves calls to the allocator, etc However, I do agree it doesn't make a measurable difference |
adriangb
approved these changes
Mar 23, 2026
adriangb
left a comment
adriangb
left a comment
Contributor
There was a problem hiding this comment.
Yeah I agree, I'm good with it either way. I've approved so if you think it's worth doing even if it doesn't show up in benches we can merge it. But I suggest we either merge or close 😄
Contributor
Author
Thank you -- let's merge it! |
Contributor
Author
de-bgunter
pushed a commit
to de-bgunter/datafusion
that referenced
this pull request
Mar 24, 2026
## Which issue does this PR close? - part of apache#19795 - follow on to apache#20234 ## Rationale for this change While reviewing apache#20234 from @AdamGS I wondered why we were creating new `RecordBatch`es to simplify expressions. I looked into it a bit and I think we can avoid a bunch of small allocations / deallocations ## What changes are included in this PR? 1. Create the dummy batch once and reuse it ## Are these changes tested? Yes by CI. I will also run benchmarks on it ## Are there any user-facing changes? No this is entirely internal
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 free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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?
simplify_const_expr#20234Rationale for this change
While reviewing #20234 from @AdamGS I wondered why we were creating new
RecordBatches to simplify expressions. I looked into it a bit and I think we can avoid a bunch of small allocations / deallocationsWhat changes are included in this PR?
Are these changes tested?
Yes by CI. I will also run benchmarks on it
Are there any user-facing changes?
No this is entirely internal