planner: mark the both side operand of NAAJ & refuse partial column substitute in projection elimination of Apply de-correlation #37117

AilinKid · 2022-08-16T03:33:23Z

Signed-off-by: AilinKid 314806019@qq.com

What problem does this PR solve?

Issue Number: close #37032

Problem Summary:
two problems mainly:
1: in the projection elimination of Apply decorrelate rule, we try to substitute any condition used in Apply with raw projected expression under this projection, after which this projection can be eliminated. But we should make sure
that all columns mapped from the schema of projection OP must be substituted atomically (all succeed or fail).

2: Consider an EQ condition converted from NOT IN OP (NAAJ), in which we used to mark the right side column of this "NA" EQ as InOperand=true. In some cases like columns substitution as we talked about above, cases like:

(a NA EQ b (marked InOperand=true ) ) where b will be substituted with like a constant string, so we got (a NA EQ "example" ), this condition will not be considered as a NA EQ condition anymore and will be moved from other condition to equal condition of join OP in the logical rule of pddResolver .

What is changed and how it works?

1: NAAJ is not just about one side, mark both as inOperand=true. so even the right has successfully been substituted with a constant string, we can also get (a (marked InOperand=true) NA EQ "example" ) as a NA EQ in the other condition.

2: make the column substitution atomically (all succeed or fail) in projection elimination of Apply decorrelation.

Check List

Tests

Unit test
Integration test
Manual test (add detailed scripts or steps below)
No code

Side effects

Performance regression: Consumes more CPU
Performance regression: Consumes more Memory
Breaking backward compatibility

Documentation

Release note

Please refer to Release Notes Language Style Guide to write a quality release note.

make the both side operand of NAAJ & refuse partial column substitute in projection elimination

… in projection elimination Signed-off-by: AilinKid <314806019@qq.com>

ti-chi-bot · 2022-08-16T03:33:24Z

[REVIEW NOTIFICATION]

This pull request has been approved by:

time-and-fate
winoros

To complete the pull request process, please ask the reviewers in the list to review by filling /cc @reviewer in the comment.
After your PR has acquired the required number of LGTMs, you can assign this pull request to the committer in the list by filling /assign @committer in the comment to help you merge this pull request.

The full list of commands accepted by this bot can be found here.

Reviewer can indicate their review by submitting an approval review.
Reviewer can cancel approval by submitting a request changes review.

sre-bot · 2022-08-16T03:34:09Z

Please follow PR Title Format:

pkg [, pkg2, pkg3]: what is changed

Or if the count of mainly changed packages are more than 3, use

*: what is changed

After you have format title, you can leave a comment /run-check_title to recheck it

ti-chi-bot · 2022-08-16T03:38:27Z

Code Coverage Details: https://codecov.io/github/pingcap/tidb/commit/f6a6d091c9f362a60addc46c87071bb229a11cf9

hawkingrei · 2022-08-16T03:47:27Z

/run-build

hawkingrei · 2022-08-16T03:56:19Z

[2022-08-16T03:55:46.307Z] planner/core/expression_rewriter.go:940:6: ineffectual assignment to larg (ineffassign)

[2022-08-16T03:55:46.307Z] planner/core/logical_plans.go:396:58: should use make([]*expression.ScalarFunction, len(p.EqualConditions)) instead (S1019)

[2022-08-16T03:55:46.307Z] planner/core/rule_decorrelate.go:174:63: should use make([]expression.Expression, len(proj.Exprs)) instead (S1019)

@AilinKid Please fix linter.

Signed-off-by: AilinKid <314806019@qq.com>

planner/core/logical_plans.go

Signed-off-by: AilinKid <314806019@qq.com>

planner/core/logical_plans.go

ti-srebot · 2022-08-24T10:06:58Z

cherry pick to release-4.0 in PR #37346