alamb commented on code in PR #8694:
URL: https://github.com/apache/arrow-rs/pull/8694#discussion_r2508177556
##########
arrow-array/src/array/byte_view_array.rs:
##########
@@ -512,18 +512,85 @@ impl<T: ByteViewType + ?Sized> GenericByteViewArray<T> {
};
}
- // 3) Allocate exactly capacity for all non-inline data
- let mut data_buf = Vec::with_capacity(total_large);
+ let (views_buf, data_blocks) = if total_large < i32::MAX as usize {
+ // fast path, the entire data fits in a single buffer
+ // 3) Allocate exactly capacity for all non-inline data
+ let mut data_buf = Vec::with_capacity(total_large);
+
+ // 4) Iterate over views and process each inline/non-inline view
+ let views_buf: Vec<u128> = (0..len)
+ .map(|i| unsafe { self.copy_view_to_buffer(i, 0, &mut
data_buf) })
+ .collect();
+ let data_block = Buffer::from_vec(data_buf);
+ let data_blocks = vec![data_block];
+ (views_buf, data_blocks)
+ } else {
+ // slow path, need to split into multiple buffers
+
+ struct GcCopyGroup {
+ total_buffer_bytes: usize,
+ total_len: usize,
+ }
+
+ impl GcCopyGroup {
+ fn new(total_buffer_bytes: u32, total_len: usize) -> Self {
+ Self {
+ total_buffer_bytes: total_buffer_bytes as usize,
+ total_len,
+ }
+ }
+ }
- // 4) Iterate over views and process each inline/non-inline view
- let views_buf: Vec<u128> = (0..len)
- .map(|i| unsafe { self.copy_view_to_buffer(i, &mut data_buf) })
- .collect();
+ let mut groups = Vec::with_capacity(total_large / (i32::MAX as
usize) + 1);
Review Comment:
I think you are referring tot his comment
- https://github.com/mapleFU/arrow-rs/pull/1#discussion_r2506878313
I didn't quite follow the concern, see
https://github.com/mapleFU/arrow-rs/pull/1#discussion_r2508175348
However that being said we can revert this part too -- I don't feel strongly
about it and I agree that the allocation likely doesn't matter on the slow path
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
