fix(cost): return full selectivity for self joins (#201)

will go after #200 Signed-off-by: Alex Chi <[email protected]> Co-authored-by: Benjamin Owad <[email protected]>
cmu-db · Oct 30, 2024 · 68c697d · 68c697d
1 parent 7b530d1
commit 68c697d
Show file tree

Hide file tree

Showing 4 changed files with 55 additions and 3 deletions.
diff --git a/optd-datafusion-repr/src/cost/base_cost/join.rs b/optd-datafusion-repr/src/cost/base_cost/join.rs
@@ -451,6 +451,10 @@ impl<
         predicate: &EqPredicate,
         past_eq_columns: &mut EqBaseTableColumnSets,
     ) -> f64 {
+        if predicate.left == predicate.right {
+            // self-join, TODO: is this correct?
+            return 1.0;
+        }
         // To find the adjustment, we need to know the selectivity of the graph before `predicate` is added.
         //
         // There are two cases: (1) adding `predicate` does not change the # of connected components, and

diff --git a/optd-datafusion-repr/src/properties/column_ref.rs b/optd-datafusion-repr/src/properties/column_ref.rs
@@ -362,7 +362,9 @@ impl PropertyBuilder<OptRelNodeTyp> for ColumnRefPropertyBuilder {
                 GroupColumnRefs::new(column_refs, child.output_correlation.clone())
             }
             // Should account for all physical join types.
-            OptRelNodeTyp::Join(join_type) | OptRelNodeTyp::RawDepJoin(join_type)  | OptRelNodeTyp::DepJoin(join_type)=> {
+            OptRelNodeTyp::Join(join_type)
+            | OptRelNodeTyp::RawDepJoin(join_type)
+            | OptRelNodeTyp::DepJoin(join_type) => {
                 // Concatenate left and right children column refs.
                 let column_refs = Self::concat_children_col_refs(&children[0..2]);
                 // Merge the equal columns of two children as input correlation.
@@ -465,12 +467,14 @@ impl PropertyBuilder<OptRelNodeTyp> for ColumnRefPropertyBuilder {
                 GroupColumnRefs::new(column_refs, correlation)
             }
             OptRelNodeTyp::Constant(_)
-            | OptRelNodeTyp::ExternColumnRef // TODO Possibly very very wrong---consult cost model team
             | OptRelNodeTyp::Func(_)
             | OptRelNodeTyp::DataType(_)
             | OptRelNodeTyp::Between
             | OptRelNodeTyp::Like
-            | OptRelNodeTyp::InList => GroupColumnRefs::new(vec![ColumnRef::Derived], None),
+            | OptRelNodeTyp::InList
+            | OptRelNodeTyp::ExternColumnRef => {
+                GroupColumnRefs::new(vec![ColumnRef::Derived], None)
+            }
             _ => unimplemented!("Unsupported rel node type {:?}", typ),
         }
     }

diff --git a/optd-sqlplannertest/tests/joins/self-join.planner.sql b/optd-sqlplannertest/tests/joins/self-join.planner.sql
@@ -0,0 +1,31 @@
+-- (no id or description)
+create table t1(t1v1 int, t1v2 int);
+create table t2(t2v1 int, t2v3 int);
+insert into t1 values (0, 0), (1, 1), (2, 2);
+insert into t2 values (0, 200), (1, 201), (2, 202);
+
+/*
+3
+3
+*/
+
+-- test self join
+select * from t1 as a, t1 as b where a.t1v1 = b.t1v1;
+
+/*
+LogicalProjection { exprs: [ #0, #1, #2, #3 ] }
+└── LogicalFilter
+    ├── cond:Eq
+    │   ├── #0
+    │   └── #2
+    └── LogicalJoin { join_type: Cross, cond: true }
+        ├── LogicalScan { table: t1 }
+        └── LogicalScan { table: t1 }
+PhysicalHashJoin { join_type: Inner, left_keys: [ #0 ], right_keys: [ #0 ] }
+├── PhysicalScan { table: t1 }
+└── PhysicalScan { table: t1 }
+0 0 0 0
+1 1 1 1
+2 2 2 2
+*/
+
diff --git a/optd-sqlplannertest/tests/joins/self-join.yml b/optd-sqlplannertest/tests/joins/self-join.yml
@@ -0,0 +1,13 @@
+- sql: |
+    create table t1(t1v1 int, t1v2 int);
+    create table t2(t2v1 int, t2v3 int);
+    insert into t1 values (0, 0), (1, 1), (2, 2);
+    insert into t2 values (0, 200), (1, 201), (2, 202);
+  tasks:
+    - execute
+- sql: |
+    select * from t1 as a, t1 as b where a.t1v1 = b.t1v1;
+  desc: test self join
+  tasks:
+    - explain:logical_optd,physical_optd
+    - execute