Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

hashjoin indexing #1728

Merged
merged 31 commits into from
Apr 27, 2023
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
31 commits
Select commit Hold shift + click to select a range
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
88 changes: 88 additions & 0 deletions enginetest/queries/queries.go
Original file line number Diff line number Diff line change
Expand Up @@ -7522,6 +7522,71 @@ SELECT * FROM my_cte;`,
{2},
},
},
{
// Original Issue: https://github.com/dolthub/dolt/issues/5714
Query: `
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

a comment tagging the original issue might be worthwhile


SELECT COUNT(*)
FROM keyless
WHERE keyless.c0 IN (

WITH RECURSIVE cte(depth, i, j) AS (
SELECT 0, T1.c0, T1.c1
FROM keyless T1
WHERE T1.c0 = 0

UNION ALL

SELECT cte.depth + 1, cte.i, T2.c1 + 1
FROM cte, keyless T2
WHERE cte.depth = T2.c0
)

SELECT U0.c0
FROM keyless U0, cte
WHERE cte.j = keyless.c0

)
ORDER BY c0;
;`,

Expected: []sql.Row{
{4},
},
},
{
// Original Issue: https://github.com/dolthub/dolt/issues/5714
// Similar, but this time the subquery is on the left
Query: `

SELECT COUNT(*)
FROM keyless
WHERE keyless.c0 IN (

WITH RECURSIVE cte(depth, i, j) AS (
SELECT 0, T1.c0, T1.c1
FROM keyless T1
WHERE T1.c0 = 0

UNION ALL

SELECT cte.depth + 1, cte.i, T2.c1 + 1
FROM cte, keyless T2
WHERE cte.depth = T2.c0
)

SELECT U0.c0
FROM cte, keyless U0
WHERE cte.j = keyless.c0

)
ORDER BY c0;
;`,

Expected: []sql.Row{
{4},
},
},
}

var KeylessQueries = []QueryTest{
Expand Down Expand Up @@ -7946,6 +8011,29 @@ var BrokenQueries = []QueryTest{
Query: "select 2000.0 / 250000000.0 * (24.0 * 6.0 * 6.25 * 10.0);",
Expected: []sql.Row{{"0.0720000000"}},
},
{
// This panics
// The non-recursive part of the UNION ALL returns too many rows, causing index out of bounds errors
// Without the join on mytable and cte, this error is caught
Query: `
WITH RECURSIVE cte(i, j) AS (
SELECT 0, 1, 2
FROM mytable

UNION ALL

SELECT *
FROM mytable, cte
WHERE cte.i = mytable.i
)
SELECT *
FROM mytable;`,
Expected: []sql.Row{
{1, "first row"},
{2, "second row"},
{3, "third row"},
},
},
}

var VersionedQueries = []QueryTest{
Expand Down
156 changes: 156 additions & 0 deletions enginetest/queries/query_plans.go
Original file line number Diff line number Diff line change
Expand Up @@ -7694,6 +7694,162 @@ With c as (
" └─ columns: [i]\n" +
"",
},
{
Query: `
SELECT COUNT(*)
FROM keyless
WHERE keyless.c0 IN (
WITH RECURSIVE cte(depth, i, j) AS (
SELECT 0, T1.c0, T1.c1
FROM keyless T1
WHERE T1.c0 = 0

UNION ALL

SELECT cte.depth + 1, cte.i, T2.c1 + 1
FROM cte, keyless T2
WHERE cte.depth = T2.c0
)

SELECT U0.c0
FROM keyless U0, cte
WHERE cte.j = keyless.c0
);`,
ExpectedPlan: "Project\n" +
" ├─ columns: [COUNT(1):0!null as COUNT(*)]\n" +
" └─ GroupBy\n" +
" ├─ select: COUNT(1 (bigint))\n" +
" ├─ group: \n" +
" └─ Filter\n" +
" ├─ InSubquery\n" +
" │ ├─ left: keyless.c0:0\n" +
" │ └─ right: Subquery\n" +
" │ ├─ cacheable: false\n" +
" │ └─ Project\n" +
" │ ├─ columns: [U0.c0:2]\n" +
" │ └─ Filter\n" +
" │ ├─ Eq\n" +
" │ │ ├─ cte.j:5\n" +
" │ │ └─ keyless.c0:0\n" +
" │ └─ CrossJoin\n" +
" │ ├─ TableAlias(U0)\n" +
" │ │ └─ Table\n" +
" │ │ ├─ name: keyless\n" +
" │ │ └─ columns: [c0]\n" +
" │ └─ SubqueryAlias\n" +
" │ ├─ name: cte\n" +
" │ ├─ outerVisibility: true\n" +
" │ ├─ cacheable: true\n" +
" │ └─ RecursiveCTE\n" +
" │ └─ Union all\n" +
" │ ├─ Project\n" +
" │ │ ├─ columns: [0 (tinyint), T1.c0:2, T1.c1:3]\n" +
" │ │ └─ Filter\n" +
" │ │ ├─ Eq\n" +
" │ │ │ ├─ T1.c0:2\n" +
" │ │ │ └─ 0 (tinyint)\n" +
" │ │ └─ TableAlias(T1)\n" +
" │ │ └─ Table\n" +
" │ │ ├─ name: keyless\n" +
" │ │ └─ columns: [c0 c1]\n" +
" │ └─ Project\n" +
" │ ├─ columns: [(cte.depth:3!null + 1 (tinyint)), cte.i:4, (T2.c1:7 + 1 (tinyint))]\n" +
" │ └─ HashJoin\n" +
" │ ├─ Eq\n" +
" │ │ ├─ cte.depth:3!null\n" +
" │ │ └─ T2.c0:6\n" +
" │ ├─ RecursiveTable(cte)\n" +
" │ └─ HashLookup\n" +
" │ ├─ source: TUPLE(cte.depth:3!null)\n" +
" │ ├─ target: TUPLE(T2.c0:2)\n" +
" │ └─ CachedResults\n" +
" │ └─ TableAlias(T2)\n" +
" │ └─ Table\n" +
" │ ├─ name: keyless\n" +
" │ └─ columns: [c0 c1]\n" +
" └─ Table\n" +
" ├─ name: keyless\n" +
" └─ columns: [c0 c1]\n" +
"",
},
{
Query: `
SELECT COUNT(*)
FROM keyless
WHERE keyless.c0 IN (
WITH RECURSIVE cte(depth, i, j) AS (
SELECT 0, T1.c0, T1.c1
FROM keyless T1
WHERE T1.c0 = 0

UNION ALL

SELECT cte.depth + 1, cte.i, T2.c1 + 1
FROM cte, keyless T2
WHERE cte.depth = T2.c0
)

SELECT U0.c0
FROM cte, keyless U0
WHERE cte.j = keyless.c0
);`,
ExpectedPlan: "Project\n" +
" ├─ columns: [COUNT(1):0!null as COUNT(*)]\n" +
" └─ GroupBy\n" +
" ├─ select: COUNT(1 (bigint))\n" +
" ├─ group: \n" +
" └─ Filter\n" +
" ├─ InSubquery\n" +
" │ ├─ left: keyless.c0:0\n" +
" │ └─ right: Subquery\n" +
" │ ├─ cacheable: false\n" +
" │ └─ Project\n" +
" │ ├─ columns: [U0.c0:5]\n" +
" │ └─ Filter\n" +
" │ ├─ Eq\n" +
" │ │ ├─ cte.j:4\n" +
" │ │ └─ keyless.c0:0\n" +
" │ └─ CrossJoin\n" +
" │ ├─ SubqueryAlias\n" +
" │ │ ├─ name: cte\n" +
" │ │ ├─ outerVisibility: true\n" +
" │ │ ├─ cacheable: true\n" +
" │ │ └─ RecursiveCTE\n" +
" │ │ └─ Union all\n" +
" │ │ ├─ Project\n" +
" │ │ │ ├─ columns: [0 (tinyint), T1.c0:2, T1.c1:3]\n" +
" │ │ │ └─ Filter\n" +
" │ │ │ ├─ Eq\n" +
" │ │ │ │ ├─ T1.c0:2\n" +
" │ │ │ │ └─ 0 (tinyint)\n" +
" │ │ │ └─ TableAlias(T1)\n" +
" │ │ │ └─ Table\n" +
" │ │ │ ├─ name: keyless\n" +
" │ │ │ └─ columns: [c0 c1]\n" +
" │ │ └─ Project\n" +
" │ │ ├─ columns: [(cte.depth:2!null + 1 (tinyint)), cte.i:3, (T2.c1:6 + 1 (tinyint))]\n" +
" │ │ └─ HashJoin\n" +
" │ │ ├─ Eq\n" +
" │ │ │ ├─ cte.depth:2!null\n" +
" │ │ │ └─ T2.c0:5\n" +
" │ │ ├─ RecursiveTable(cte)\n" +
" │ │ └─ HashLookup\n" +
" │ │ ├─ source: TUPLE(cte.depth:2!null)\n" +
" │ │ ├─ target: TUPLE(T2.c0:2)\n" +
" │ │ └─ CachedResults\n" +
" │ │ └─ TableAlias(T2)\n" +
" │ │ └─ Table\n" +
" │ │ ├─ name: keyless\n" +
" │ │ └─ columns: [c0 c1]\n" +
" │ └─ TableAlias(U0)\n" +
" │ └─ Table\n" +
" │ ├─ name: keyless\n" +
" │ └─ columns: [c0]\n" +
" └─ Table\n" +
" ├─ name: keyless\n" +
" └─ columns: [c0 c1]\n" +
"",
},
}

// QueryPlanTODOs are queries where the query planner produces a correct (results) but suboptimal plan.
Expand Down
1 change: 0 additions & 1 deletion sql/analyzer/analyzer.go
Original file line number Diff line number Diff line change
Expand Up @@ -528,7 +528,6 @@ func postPrepareRuleSelector(id RuleId) bool {
stripTableNameInDefaultsId,
resolvePreparedInsertId,
finalizeSubqueriesId,
finalizeUnionsId,

// DefaultValidationRules
validateResolvedId,
Expand Down
6 changes: 5 additions & 1 deletion sql/analyzer/exec_builder.go
Original file line number Diff line number Diff line change
Expand Up @@ -180,7 +180,11 @@ func (b *ExecBuilder) buildHashJoin(j *hashJoin, input sql.Schema, children ...s
if err != nil {
return nil, err
}
outerAttrs, err := b.buildFilters(j.g.m.scope, j.right.relProps.OutputCols(), expression.Tuple(j.outerAttrs))
tmpScope := j.g.m.scope
if tmpScope != nil {
tmpScope = tmpScope.newScopeNoJoin()
}
outerAttrs, err := b.buildFilters(tmpScope, j.right.relProps.OutputCols(), expression.Tuple(j.outerAttrs))
if err != nil {
return nil, err
}
Expand Down
1 change: 1 addition & 0 deletions sql/analyzer/indexed_joins.go
Original file line number Diff line number Diff line change
Expand Up @@ -117,6 +117,7 @@ func inOrderReplanJoin(

// two different base cases, depending on whether we reorder or not
if reorder {
scope.SetJoin(true)
ret, err := replanJoin(ctx, j, a, scope)
if err != nil {
return nil, transform.SameTree, fmt.Errorf("failed to replan join: %w", err)
Expand Down
3 changes: 2 additions & 1 deletion sql/analyzer/pushdown.go
Original file line number Diff line number Diff line change
Expand Up @@ -953,7 +953,8 @@ func convertIsNullForIndexes(ctx *sql.Context, e sql.Expression) sql.Expression
// pushdownFixIndices fixes field indices for non-join expressions (replanJoin
// is responsible for join filters and conditions.)
func pushdownFixIndices(a *Analyzer, n sql.Node, scope *Scope) (sql.Node, transform.TreeIdentity, error) {
if _, ok := n.(*plan.JoinNode); ok {
switch n := n.(type) {
case *plan.JoinNode, *plan.HashLookup:
return n, transform.SameTree, nil
}
return FixFieldIndexesForExpressions(a, n, scope)
Expand Down
Loading