You are viewing a plain text version of this content. The canonical link for it is here.
Posted to github@datafusion.apache.org by "berkaysynnada (via GitHub)" <gi...@apache.org> on 2025/07/24 08:26:47 UTC
[PR] Fix Partial Sort Get Slice Point Between Batches [datafusion]
berkaysynnada opened a new pull request, #16881:
URL: https://github.com/apache/datafusion/pull/16881
## Which issue does this PR close?
<!--
We generally require a GitHub issue to be filed for all bug fixes and enhancements and this helps us generate change logs for our releases. You can link an issue to this PR using the GitHub syntax. For example `Closes #123` indicates that this PR will close issue #123.
-->
- Closes #.
## Rationale for this change
<!--
Why are you proposing this change? If this is already explained clearly in the issue then this section is not needed.
Explaining clearly why changes are proposed helps reviewers understand your changes and offer better suggestions for fixes.
-->
PartialSortExec had a missing functionality where it failed to detect slice points when consecutive batches contained homogeneous sort key values (e.g., first batch: all rows with a=1, second batch: all rows with a=2). The get_slice_point method only checked within individual batches, missing cross-batch transitions.
## What changes are included in this PR?
<!--
There is no need to duplicate the description in the issue here but it is sometimes worth providing a summary of the individual changes in this PR.
-->
* Modified the batching logic to accumulate incoming batches into a single buffer before checking for slice points
* Added proper handling of fetch limits to avoid processing after the limit is reached
## Are these changes tested?
<!--
We typically require tests for all PRs in order to:
1. Prevent the code from being accidentally broken by subsequent changes
2. Serve as another way to document the expected behavior of the code
If tests are not included in your PR, please explain why (for example, are they covered by existing tests)?
-->
Yes
## Are there any user-facing changes?
<!--
If there are user-facing changes then we may require documentation to be updated before approving the PR.
-->
<!--
If there are any breaking changes to public APIs, please add the `api change` label.
-->
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] Fix Partial Sort Get Slice Point Between Batches [datafusion]
Posted by "berkaysynnada (via GitHub)" <gi...@apache.org>.
berkaysynnada commented on code in PR #16881:
URL: https://github.com/apache/datafusion/pull/16881#discussion_r2230512203
##########
datafusion/physical-plan/src/sorts/partial_sort.rs:
##########
@@ -375,34 +375,52 @@ impl PartialSortStream {
return Poll::Ready(None);
}
loop {
+ // Check if we've already reached the fetch limit
+ if self.fetch == Some(0) {
+ self.is_closed = true;
+ return Poll::Ready(None);
+ }
+
return Poll::Ready(match ready!(self.input.poll_next_unpin(cx)) {
Some(Ok(batch)) => {
- if let Some(slice_point) =
- self.get_slice_point(self.common_prefix_length, &batch)?
+ // Merge new batch into in_mem_batch
+ self.in_mem_batch = concat_batches(
+ &self.schema(),
+ &[self.in_mem_batch.clone(), batch],
+ )?;
+
+ // Check if we have a slice point, otherwise keep accumulating in `self.in_mem_batch`.
+ if let Some(slice_point) = self
+ .get_slice_point(self.common_prefix_length, &self.in_mem_batch)?
{
- self.in_mem_batches.push(batch.slice(0, slice_point));
- let remaining_batch =
- batch.slice(slice_point, batch.num_rows() - slice_point);
- // Extract the sorted batch
- let sorted_batch = self.sort_in_mem_batches();
- // Refill with the remaining batch
- self.in_mem_batches.push(remaining_batch);
-
- debug_assert!(sorted_batch
- .as_ref()
- .map(|batch| batch.num_rows() > 0)
- .unwrap_or(true));
- Some(sorted_batch)
+ let sorted = self.in_mem_batch.slice(0, slice_point);
+ self.in_mem_batch = self.in_mem_batch.slice(
+ slice_point,
+ self.in_mem_batch.num_rows() - slice_point,
+ );
+ let sorted_batch = sort_batch(&sorted, &self.expr, self.fetch)?;
+ if let Some(fetch) = self.fetch.as_mut() {
+ *fetch -= sorted_batch.num_rows();
+ // If we've reached the fetch limit, close the stream
+ if *fetch == 0 {
Review Comment:
😳
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] Fix Partial Sort Get Slice Point Between Batches [datafusion]
Posted by "alamb (via GitHub)" <gi...@apache.org>.
alamb commented on code in PR #16881:
URL: https://github.com/apache/datafusion/pull/16881#discussion_r2229200729
##########
datafusion/physical-plan/src/sorts/partial_sort.rs:
##########
@@ -375,34 +375,52 @@ impl PartialSortStream {
return Poll::Ready(None);
}
loop {
+ // Check if we've already reached the fetch limit
+ if self.fetch == Some(0) {
+ self.is_closed = true;
+ return Poll::Ready(None);
+ }
+
return Poll::Ready(match ready!(self.input.poll_next_unpin(cx)) {
Some(Ok(batch)) => {
- if let Some(slice_point) =
- self.get_slice_point(self.common_prefix_length, &batch)?
+ // Merge new batch into in_mem_batch
+ self.in_mem_batch = concat_batches(
+ &self.schema(),
+ &[self.in_mem_batch.clone(), batch],
+ )?;
+
+ // Check if we have a slice point, otherwise keep accumulating in `self.in_mem_batch`.
+ if let Some(slice_point) = self
+ .get_slice_point(self.common_prefix_length, &self.in_mem_batch)?
{
- self.in_mem_batches.push(batch.slice(0, slice_point));
- let remaining_batch =
- batch.slice(slice_point, batch.num_rows() - slice_point);
- // Extract the sorted batch
- let sorted_batch = self.sort_in_mem_batches();
- // Refill with the remaining batch
- self.in_mem_batches.push(remaining_batch);
-
- debug_assert!(sorted_batch
- .as_ref()
- .map(|batch| batch.num_rows() > 0)
- .unwrap_or(true));
- Some(sorted_batch)
+ let sorted = self.in_mem_batch.slice(0, slice_point);
+ self.in_mem_batch = self.in_mem_batch.slice(
+ slice_point,
+ self.in_mem_batch.num_rows() - slice_point,
+ );
+ let sorted_batch = sort_batch(&sorted, &self.expr, self.fetch)?;
+ if let Some(fetch) = self.fetch.as_mut() {
+ *fetch -= sorted_batch.num_rows();
+ // If we've reached the fetch limit, close the stream
+ if *fetch == 0 {
+ self.is_closed = true;
+ }
+ }
+
+ if sorted_batch.num_rows() > 0 {
+ Some(Ok(sorted_batch))
+ } else {
+ continue;
Review Comment:
stylistically I found this `continue` in the middle of a `return` statement somewhat confusing
I wonder if we could reduce the indent of this code / make it clearer when control returned by moving the `return` to the locations where it is used
like
```rust
if sorted_batch.num_rows() > 0 {
return Poll::Ready(Some(Ok(sorted_batch)))
} else {
continue;
```
instead of
```rust
return Poll::Ready(....
if sorted_batch.num_rows() > 0 {
Some(Ok(sorted_batch))
} else {
continue;
```
##########
datafusion/physical-plan/src/sorts/partial_sort.rs:
##########
@@ -375,34 +375,52 @@ impl PartialSortStream {
return Poll::Ready(None);
}
loop {
+ // Check if we've already reached the fetch limit
+ if self.fetch == Some(0) {
+ self.is_closed = true;
+ return Poll::Ready(None);
+ }
+
return Poll::Ready(match ready!(self.input.poll_next_unpin(cx)) {
Some(Ok(batch)) => {
- if let Some(slice_point) =
- self.get_slice_point(self.common_prefix_length, &batch)?
+ // Merge new batch into in_mem_batch
+ self.in_mem_batch = concat_batches(
+ &self.schema(),
+ &[self.in_mem_batch.clone(), batch],
+ )?;
+
+ // Check if we have a slice point, otherwise keep accumulating in `self.in_mem_batch`.
+ if let Some(slice_point) = self
+ .get_slice_point(self.common_prefix_length, &self.in_mem_batch)?
{
- self.in_mem_batches.push(batch.slice(0, slice_point));
- let remaining_batch =
- batch.slice(slice_point, batch.num_rows() - slice_point);
- // Extract the sorted batch
- let sorted_batch = self.sort_in_mem_batches();
- // Refill with the remaining batch
- self.in_mem_batches.push(remaining_batch);
-
- debug_assert!(sorted_batch
- .as_ref()
- .map(|batch| batch.num_rows() > 0)
- .unwrap_or(true));
- Some(sorted_batch)
+ let sorted = self.in_mem_batch.slice(0, slice_point);
+ self.in_mem_batch = self.in_mem_batch.slice(
+ slice_point,
+ self.in_mem_batch.num_rows() - slice_point,
+ );
+ let sorted_batch = sort_batch(&sorted, &self.expr, self.fetch)?;
+ if let Some(fetch) = self.fetch.as_mut() {
+ *fetch -= sorted_batch.num_rows();
+ // If we've reached the fetch limit, close the stream
+ if *fetch == 0 {
Review Comment:
Is this `if` redundant with the condition at the top of the loop?
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] Fix Partial Sort Get Slice Point Between Batches [datafusion]
Posted by "berkaysynnada (via GitHub)" <gi...@apache.org>.
berkaysynnada commented on PR #16881:
URL: https://github.com/apache/datafusion/pull/16881#issuecomment-3117295272
> The only thing I would also like to see is a SQL level test. However, I was not able to write one as I can't seem to get `PartialSortExec` to appear in the tests. I filed a ticket about this:
Can you check my last commit? I think it's what you want but I'm not sure about the location
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] Fix Partial Sort Get Slice Point Between Batches [datafusion]
Posted by "berkaysynnada (via GitHub)" <gi...@apache.org>.
berkaysynnada merged PR #16881:
URL: https://github.com/apache/datafusion/pull/16881
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org