This is an automated email from the ASF dual-hosted git repository.
Jefffrey pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/arrow-rs.git
The following commit(s) were added to refs/heads/main by this push:
new 195aac7c77 Replace BufferBuilder with Vec in fixed-size binary take
(#10773)
195aac7c77 is described below
commit 195aac7c776f867556545614c79d29da62a0731e
Author: kowanietz <[email protected]>
AuthorDate: Sat Aug 22 08:13:30 2026 +0400
Replace BufferBuilder with Vec in fixed-size binary take (#10773)
# Which issue does this PR close?
- part of #10245
# Rationale for this change
Replacing `BufferBuilder` with `Vec` improves the performance of the
dynamic fixed-size binary take path.
# What changes are included in this PR?
Replaces the remaining `BufferBuilder<u8>` in
`take_fixed_size_binary_buffer_dynamic_length` with a `Vec<u8>`.
# Are these changes tested?
All tests pass:
- `cargo fmt --all -- --check`
- `cargo clippy -p arrow-select --all-targets --all-features --no-deps
-- -D warnings`
- `cargo test -p arrow-select --all-features`
local benchmark results:
| Benchmark | Before | After | Improvement |
|---|---:|---:|---:|
| Size 12, no null values | 5.0871 µs | 4.5750 µs | 9.99% |
| Size 12, 50% null values | 6.2656 µs | 5.7102 µs | 8.88% |
# Are there any user-facing changes?
No.
---
arrow-select/src/take.rs | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/arrow-select/src/take.rs b/arrow-select/src/take.rs
index d86d888adc..36cdb81e76 100644
--- a/arrow-select/src/take.rs
+++ b/arrow-select/src/take.rs
@@ -21,7 +21,7 @@ use std::fmt::Display;
use std::mem::ManuallyDrop;
use std::sync::Arc;
-use arrow_array::builder::{BufferBuilder, UInt32Builder};
+use arrow_array::builder::UInt32Builder;
use arrow_array::cast::AsArray;
use arrow_array::types::*;
use arrow_array::*;
@@ -830,7 +830,7 @@ fn take_fixed_size_binary<IndexType: ArrowPrimitiveType>(
size_usize: usize,
) -> Buffer {
let values_buffer = values.values().as_slice();
- let mut values_buffer_builder = BufferBuilder::new(indices.len() *
size_usize);
+ let mut output = Vec::with_capacity(indices.len() * size_usize);
if indices.null_count() == 0 {
let array_iter = indices.values().iter().map(|idx| {
@@ -838,7 +838,7 @@ fn take_fixed_size_binary<IndexType: ArrowPrimitiveType>(
&values_buffer[offset..offset + size_usize]
});
for slice in array_iter {
- values_buffer_builder.append_slice(slice);
+ output.extend_from_slice(slice);
}
} else {
// The indices nullability cannot be ignored here because the
values buffer may contain
@@ -851,13 +851,13 @@ fn take_fixed_size_binary<IndexType: ArrowPrimitiveType>(
});
for slice in array_iter {
match slice {
- None => values_buffer_builder.append_n(size_usize, 0),
- Some(slice) => values_buffer_builder.append_slice(slice),
+ None => output.resize(output.len() + size_usize, 0),
+ Some(slice) => output.extend_from_slice(slice),
}
}
}
- values_buffer_builder.finish()
+ output.into()
}
}
