This is an automated email from the ASF dual-hosted git repository.
alamb 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 98deaf182e Replace BufferBuilder with Vec in cast_byte_container
(#10867)
98deaf182e is described below
commit 98deaf182e0a0c2da8c7621d44d598ef1ecb7e43
Author: Xinyao Zhang <[email protected]>
AuthorDate: Wed Aug 26 17:47:34 2026 -0400
Replace BufferBuilder with Vec in cast_byte_container (#10867)
# Which issue does this PR close?
- Part of #10245.
# Rationale for this change
This replaces another remaining `BufferBuilder` callsite listed in
#10245. Using `Vec` for the converted offsets avoids the builder
overhead.
# What changes are included in this PR?
`cast_byte_container` now collects converted offsets in a
`Vec<TO::Offset>` and converts it directly into an Arrow `Buffer`.
# Are these changes tested?
Yes:
- `cargo fmt --all -- --check`
- `cargo test -p arrow-cast` — 375 unit tests and 11 doc tests passed
- `cargo clippy -p arrow-cast --all-targets --all-features -- -D
warnings`
Existing string and binary cast tests cover the changed path, so no new
tests were added.
A temporary local Criterion benchmark for a 512-row `Utf8` to
`LargeUtf8` cast produced:
| | Time |
| --- | ---: |
| `main` | 504.35 ns |
| This PR | 403.76 ns |
Criterion reported a 19.9% improvement.
# Are there any user-facing changes?
No.
# AI assistance
Codex was used to identify the callsite, prepare the focused refactor,
run validation and benchmarking, and draft this description.
---
arrow-cast/src/cast/mod.rs | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/arrow-cast/src/cast/mod.rs b/arrow-cast/src/cast/mod.rs
index b6e03bdef3..1d95d2dd36 100644
--- a/arrow-cast/src/cast/mod.rs
+++ b/arrow-cast/src/cast/mod.rs
@@ -65,7 +65,7 @@ use crate::parse::{
string_to_datetime,
};
use arrow_array::{builder::*, cast::*, temporal_conversions::*, timezone::Tz,
types::*, *};
-use arrow_buffer::{ArrowNativeType, OffsetBuffer, i256};
+use arrow_buffer::{ArrowNativeType, Buffer, OffsetBuffer, i256};
use arrow_data::ArrayData;
use arrow_data::transform::MutableArrayData;
use arrow_schema::*;
@@ -2832,7 +2832,7 @@ where
let str_values_buf = data.buffers()[1].clone();
let offsets = data.buffers()[0].typed_data::<FROM::Offset>();
- let mut offset_builder = BufferBuilder::<TO::Offset>::new(offsets.len());
+ let mut cast_offsets = Vec::<TO::Offset>::with_capacity(offsets.len());
offsets
.iter()
.try_for_each::<_, Result<_, ArrowError>>(|offset| {
@@ -2846,11 +2846,11 @@ where
TO::PREFIX
))
})?;
- offset_builder.append(offset);
+ cast_offsets.push(offset);
Ok(())
})?;
- let offset_buffer = offset_builder.finish();
+ let offset_buffer = Buffer::from_vec(cast_offsets);
let dtype = TO::DATA_TYPE;
