lyne7-sc opened a new pull request, #25413:
URL: https://github.com/apache/datafusion/pull/25413

   ## 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.  
   
   Please explain the problem you are trying to solve in terms of the 
user-visible
   behavior, rather than the implementation.
   
   For example, "The code in `foo.rs` doesn't handle nulls" is a symptom of the
   implementation. "COUNT(DISTINCT) returns wrong results when the column 
contains
   nulls" is the user-visible problem.
   -->
   
   For `RightSemi` and `RightAnti` nested loop joins, each right row only needs 
one match to determine whether it belongs in the result. Currently, matched 
right rows continue participating in comparisons with subsequent left rows.
   
   Skipping these comparisons can significantly reduce execution time when 
matches are found early.
   
   ## 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.
   -->
   
   - Add `RightExistenceProbe` to track matches and right-side candidates.
   - Compact the right-side filter columns when at least half the candidates 
have matched and more left rows remain.
   - Stop probing the current right batch when all its rows have matched.
   - Preserve matches in original right-row coordinates for output and spill 
handling.
   
   ## What is the testing strategy for this PR?
   
   <!--
   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
   
   Briefly describe how this PR is tested, and point to the specific tests you 
added. For example: 'This new feature is covered by the `sqllogictest` cases 
added in `foo.slt`'.
   
   If this PR does not add tests, explain why. For example, if the change is 
already covered by existing tests, please mention it.
   
   You should also check the `codecov` bot reply on this PR to confirm the 
changed code is exercised.
   -->
   
   - Add unit tests for match tracking and candidate compaction.
   - Add SQL coverage in `nested_loop_join_existence.slt` for 
RightSemi/RightAnti joins, including NULLs, repeated matches, and multiple left 
rows per probe.
   - Extend `nested_loop_join_spill.slt` to cover RightSemi/RightAnti joins 
across spilled left chunks.
   
   ## 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.
   -->
   
   No. This is an internal performance optimization.
   
   ## Benchmark results
   
   | Case | Scenario | main | This PR | Time change |
   |---|---|---:|---:|---:|
   | Q15 | Right Semi, early matches | 307.659 | 0.365 | -99.88% |
   | Q16 | Right Anti, early matches | 306.225 | 0.338 | -99.89% |
   | Q18 | Right Semi, no matches | 7.964 | 7.728 | -2.95% |
   | Q19 | Right Anti, no matches | 8.100 | 7.770 | -4.07% |
   | Q20 | Right Semi, staged compaction | 56.082 | 7.182 | -87.19% |
   | Q21 | Right Anti, staged compaction | 56.057 | 7.115 | -87.31% |
   | Q22 | Right Semi, repeated 40% matches (4,096 left × 8,192 right) | 12.185 
| 12.410 | +1.84% |
   | Q23 | Right Anti, final-left-row matches with wide Utf8 | 0.305125 | 
0.309042 | +1.28% |


-- 
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]

Reply via email to