Introduce UNWIND_DEDUP operator #171
Open
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
Problem: When UNWIND produces duplicate values in a single batch, MERGE creates multiple nodes instead of reusing the same node. For example, incorrectly creates two separate nodes because the hash join probe doesn't see nodes created earlier in the same batch. Solution: 1. UNWIND_DEDUP Operator - Created LogicalUnwindDeduplicate and PhysicalUnwindDedup operators - Deduplicates UNWIND output by hash before MERGE, ensuring each unique value appears once - Fixes the core deduplication issue for MERGE patterns 2. Optimizer Integration - Created UnwindDedupOptimizer that inserts UNWIND_DEDUP before MERGE's hash join probe side - Triggers when MERGE probe child is a HASH_JOIN whose probe child is UNWIND 3. Constant Folding for Deterministic Inputs - Added constant folding in Planner::appendUnwind() for deterministic UNWIND expressions - Uses ConstantExpressionVisitor::isConstant() to detect foldable expressions - Non-deterministic functions (gen_random_uuid, rand, nextval) are NOT folded 4. Test Coverage (issue_170) - → 1 row (1) - → 1 node - → 2 (non-deterministic, not folded)
…perators The UNWIND_DEDUP optimizer was only detecting UNWIND when it was the direct child of HASH_JOIN in the query plan. However, when using gen_random_uuid() or other default expressions, the planner may insert intermediate operators like FLATTEN between HASH_JOIN and UNWIND, causing the pattern match to fail. This fix adds recursive tree traversal to find UNWIND nodes even when they are nested deeper in the probe child subtree. When found, the optimizer now correctly wraps the entire probe child with UNWIND_DEDUP to deduplicate values before MERGE processing. Fixes issue where MERGE with duplicate UNWIND values (e.g., UNWIND [1, 1]) would create multiple nodes with different gen_random_uuid() values instead of recognizing them as duplicates and creating only one node.
This needs deeper thought. We need a clearer language spec on when MERGE is expected to return multiple rows vs optimize it away. Previous test failures: e2e_test_transaction~create_node~merge_tinysnb_checkpoint.MergeNodeWithOnMatchAndOnCreate (Failed) e2e_test_transaction~create_node~merge_tinysnb_checkpoint.MergeRelWithOnMatch (Failed) e2e_test_dml_node~merge~merge_tinysnb.MergeNodeWithOnMatchAndOnCreate (Failed) e2e_test_dml_node~merge~merge_tinysnb.MergeRelWithOnMatch (Failed) e2e_test_dml_node~merge~merge_tinysnb.MergeNodeWithOnMatch (Failed)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Fixes: #170
Problem:
When UNWIND produces duplicate values in a single batch, MERGE creates multiple nodes instead of reusing the same node. For example, incorrectly creates two separate nodes because the hash join probe doesn't see nodes created earlier in the same batch.
Solution:
UNWIND_DEDUP Operator
Optimizer Integration
Constant Folding for Deterministic Inputs
Test Coverage (issue_170)