fix: MergeJoin is buggy with RIGHT OUTER JOINs where NULLs are present in the keys #13039
Conversation
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
facebook-github-bot
added
the
CLA Signed
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D73077550 |
✅ Deploy Preview for meta-velox canceled.
|
kevinwilfong
force-pushed
the
export-D73077550
branch
from
efe7626 to
fbaaf1d
Compare
kevinwilfong
added a commit
to kevinwilfong/velox
that referenced
this pull request
Apr 16, 2025
…t in the keys (facebookincubator#13039) Summary: While working on another change I discovered that JoinFuzzer does not test MergeJoin with RIGHT OUTER JOINs. Enabling it locally (I'll publish a change to enable it separately) I discovered a bug. While trying to add a unit test for it, I discovered a few more. They fit into two classes: 1) Skipping over rows on the right side with NULL keys. This is the correct thing to do for INNER and LEFT OUTER JOINs but we need to output misses for these rows in RIGHT OUTER JOINs (they can't hit given our NULL semantics). 2) Writing off the end of the output buffer trying to write out this misses. We need to make sure the size of output_ hasn't yet reached outputBatchSize_ before writing misses to it. This diff fixes the bugs I found and adds unit tests covering NULL keys (I didn't see any prior to this change). Differential Revision: D73077550
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D73077550 |
xiaoxmeng
approved these changes
Apr 16, 2025
…t in the keys (facebookincubator#13039) Summary: While working on another change I discovered that JoinFuzzer does not test MergeJoin with RIGHT OUTER JOINs. Enabling it locally (I'll publish a change to enable it separately) I discovered a bug. While trying to add a unit test for it, I discovered a few more. They fit into two classes: 1) Skipping over rows on the right side with NULL keys. This is the correct thing to do for INNER and LEFT OUTER JOINs but we need to output misses for these rows in RIGHT OUTER JOINs (they can't hit given our NULL semantics). 2) Writing off the end of the output buffer trying to write out this misses. We need to make sure the size of output_ hasn't yet reached outputBatchSize_ before writing misses to it. This diff fixes the bugs I found and adds unit tests covering NULL keys (I didn't see any prior to this change). Reviewed By: xiaoxmeng Differential Revision: D73077550
kevinwilfong
force-pushed
the
export-D73077550
branch
from
fbaaf1d to
f5ab1bb
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D73077550 |
Contributor
|
This pull request has been merged in 53ea8a6. |
Collaborator
|
@kevinwilfong @xiaoxmeng would you please also help to review below two merge join fixes? There were discovered in Spark/Gluten tests and fixed by @JkSelf Thanks, -yuan |
Contributor
|
Right outer join is indeed not very stable. We discussed that ideally we should only implement a left outer join, and for right outer join we just swap the inputs and run the same left outer join path. So at least we would have a single code path to test. |
Collaborator
|
@pedroerp @xiaoxmeng @kevinwilfong |
zhanglistar
pushed a commit
to bigo-sg/velox
that referenced
this pull request
Apr 22, 2025
…t in the keys (facebookincubator#13039) Summary: Pull Request resolved: facebookincubator#13039 While working on another change I discovered that JoinFuzzer does not test MergeJoin with RIGHT OUTER JOINs. Enabling it locally (I'll publish a change to enable it separately) I discovered a bug. While trying to add a unit test for it, I discovered a few more. They fit into two classes: 1) Skipping over rows on the right side with NULL keys. This is the correct thing to do for INNER and LEFT OUTER JOINs but we need to output misses for these rows in RIGHT OUTER JOINs (they can't hit given our NULL semantics). 2) Writing off the end of the output buffer trying to write out this misses. We need to make sure the size of output_ hasn't yet reached outputBatchSize_ before writing misses to it. This diff fixes the bugs I found and adds unit tests covering NULL keys (I didn't see any prior to this change). Reviewed By: xiaoxmeng Differential Revision: D73077550 fbshipit-source-id: 82d914d38835b51f52676cfa2317fdce164ee0fc
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
6 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.
Summary:
While working on another change I discovered that JoinFuzzer does not test MergeJoin with
RIGHT OUTER JOINs. Enabling it locally (I'll publish a change to enable it separately) I
discovered a bug. While trying to add a unit test for it, I discovered a few more.
They fit into two classes:
INNER and LEFT OUTER JOINs but we need to output misses for these rows in RIGHT OUTER
JOINs (they can't hit given our NULL semantics).
sure the size of output_ hasn't yet reached outputBatchSize_ before writing misses to it.
This diff fixes the bugs I found and adds unit tests covering NULL keys (I didn't see any prior
to this change).
Differential Revision: D73077550