fix statistics propagation for anti-joins on empty tables#17439
Merged
Mytherin merged 2 commits intoduckdb:v1.3-ossivalisfrom Jun 30, 2025
Merged
fix statistics propagation for anti-joins on empty tables#17439Mytherin merged 2 commits intoduckdb:v1.3-ossivalisfrom
Mytherin merged 2 commits intoduckdb:v1.3-ossivalisfrom
Conversation
68ab1ce to
22fabcc
Compare
|
Thanks for your fixing! |
Damon07
reviewed
May 12, 2025
22fabcc to
5891777
Compare
|
Dear Developers, thanks for your fixing! Could you please review this pull request and merge it? Thanks! I am currently working on extensively testing DuckDB, and I also want to express my gratitude to your fixing! |
1 task
|
also chiming in with enthusiasm for this being merged, and thanks to the dev who fixed it!! |
Tmonster
approved these changes
Jun 10, 2025
Contributor
Tmonster
left a comment
There was a problem hiding this comment.
Looks good to me. I think you can change the base branch to v1.3-Ossivalis so that this gets merged into the next bug fix release
Contributor
Author
|
Thanks for the review, changed the base branch to v1.3-Ossivalis! |
Contributor
Collaborator
|
Thanks! |
github-actions bot
pushed a commit
to duckdb/duckdb-r
that referenced
this pull request
Jun 30, 2025
fix statistics propagation for anti-joins on empty tables (duckdb/duckdb#17439) Run Python workflow against both Python 3.9 and 3.13 on PR to ensure … (duckdb/duckdb#18080) Main.yml: Move very long job from debug to release with `-DDEBUG` and `FORCE_ASSERT` (duckdb/duckdb#18081)
github-actions bot
added a commit
to duckdb/duckdb-r
that referenced
this pull request
Jun 30, 2025
fix statistics propagation for anti-joins on empty tables (duckdb/duckdb#17439) Run Python workflow against both Python 3.9 and 3.13 on PR to ensure … (duckdb/duckdb#18080) Main.yml: Move very long job from debug to release with `-DDEBUG` and `FORCE_ASSERT` (duckdb/duckdb#18081) Co-authored-by: krlmlr <krlmlr@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
7 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #17417
This issue appears to actually be a regression of issue #9308 (with the existing test for that issue not picking up the regression as the optimizer no longer makes an explicit ANTI-JOIN for the WHERE NOT EXISTS query).
While the fix in #9654 could be re-applied (and if this PR is deemed not suitable, should be) to remove the pruning optimization altogether for ANTI-JOINS with always true filters, this PR aims to keep the optimization and handle the case where the join is on an empty result set.
In particular, this PR adds a helper function to BaseStatistics to determine whether the stats match an empty result set (using the min/max set by CreateEmpty) and correctly returns the LHS of the ANTI-JOIN if the RHS is empty (and keeps the current behaviour of returning an empty result otherwise)
While this function is only currently used for NumericStats, I've also implemented it for the other BaseStatistics implementations, to avoid this being regressed again in future if statistics propagation (and in particular, filter pruning) occur for non-numeric types.