ORCA: fall back to Postgres planner for KNN ORDER BY queries#1653
Open
yjhjstz wants to merge 1 commit intoapache:mainfrom
Open
ORCA: fall back to Postgres planner for KNN ORDER BY queries#1653yjhjstz wants to merge 1 commit intoapache:mainfrom
yjhjstz wants to merge 1 commit intoapache:mainfrom
Conversation
my-ship-it
approved these changes
Apr 1, 2026
my-ship-it
reviewed
Apr 1, 2026
| * (not wrapped in a function or other expression). | ||
| */ | ||
| static bool | ||
| has_plain_var_arg(OpExpr *op) |
Contributor
There was a problem hiding this comment.
The check requires a direct Var node. A RelabelType wrapping a Var (common with type coercions) would not match:
-- If the column is int2 but the operator expects int4,
-- the Var gets wrapped in RelabelType
ORDER BY int2_col <-> 42
Member
Author
There was a problem hiding this comment.
fixed by strip_implicit_coercions
Contributor
|
will try to take a look soon. Changes looks sane |
ORCA is unaware of amcanorderbyop, so it plans "ORDER BY col <-> val" queries with a full Seq Scan + Sort instead of a native KNN ordered index scan. Detect this pattern by checking whether any ORDER BY target is an operator with amoppurpose = AMOP_ORDER in pg_amop and at least one direct Var argument, then raise ExmiQuery2DXLUnsupportedFeature to hand the query off to the Postgres planner, which generates an efficient Index Only Scan with native KNN ordering. Queries where the ordering operator's arguments are entirely computed expressions (e.g. circle(col,1) <-> point(0,0)) are excluded from the fallback to avoid lossy-distance errors in index-only scans.
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
3 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.
ORCA is unaware of amcanorderbyop, so it plans "ORDER BY col <-> val"
queries with a full Seq Scan + Sort instead of a native KNN ordered
index scan. Detect this pattern by checking whether any ORDER BY
target is an operator with amoppurpose = AMOP_ORDER in pg_amop and at
least one direct Var argument, then raise ExmiQuery2DXLUnsupportedFeature
to hand the query off to the Postgres planner, which generates an
efficient Index Only Scan with native KNN ordering.
Queries where the ordering operator's arguments are entirely computed
expressions (e.g. circle(col,1) <-> point(0,0)) are excluded from the
fallback to avoid lossy-distance errors in index-only scans.
Fixes #ISSUE_Number
What does this PR do?
Type of Change
Breaking Changes
Test Plan
make installcheckmake -C src/test installcheck-cbdb-parallelImpact
Performance:
User-facing changes:
Dependencies:
Checklist
Additional Context
CI Skip Instructions