comphead commented on code in PR #3059:
URL: https://github.com/apache/datafusion-comet/pull/3059#discussion_r2677351186
##########
native/core/src/execution/shuffle/row.rs:
##########
@@ -776,16 +776,27 @@ pub fn process_sorted_row_partition(
None
};
+ // Create builders once and reuse them across batches.
+ // After finish() is called, builders are reset and can be reused.
+ let mut data_builders: Vec<Box<dyn ArrayBuilder>> = vec![];
+ schema.iter().try_for_each(|dt| {
+ make_builders(dt, batch_size, prefer_dictionary_ratio)
+ .map(|builder| data_builders.push(builder))?;
+ Ok::<(), CometError>(())
+ })?;
+
+ // Open the output file once and reuse it across batches
+ let mut output_data = OpenOptions::new()
+ .create(true)
+ .append(true)
+ .open(&output_path)?;
+
+ // Reusable buffer for serialized batch data
+ let mut frozen: Vec<u8> = Vec::with_capacity(batch_size * 1024);
Review Comment:
it looks good for me, however, what is 1024? are we considering 1KB is the a
single value size?
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
