[CALCITE-6272] Improved LogicalPlan representation of distinct keyword

• Type: Improvement
• Status: Closed
• Priority: Minor
• Resolution: Not A Problem
• Affects Version/s: 1.36.0
• Fix Version/s: 1.37.0
• Component/s: core
• Assignee: Unassigned
• Reporter: Caican Cai
• Created: 17/Feb/24 13:17
• Updated: 20/Feb/24 06:41
• Resolved: 20/Feb/24 06:41

Description

In the ToLogicalConvert test, the LogicalPlan parsed with or without the distinct keyword is the same. I think this is an improvement.

 

@Test void testdistnct() {
  // Equivalent SQL:
  //   SELECT DISTINCT *
  //   FROM emp
  final RelBuilder builder = builder();
  final RelNode rel =
      builder.scan("EMP")
          .distinct()
          .build();
  String expectedPhysical = ""
      + "EnumerableTableScan(table=[[scott, EMP]])\n";
  String expectedLogical = ""
      + "LogicalTableScan(table=[[scott, EMP]])\n";
  verify(rel, expectedPhysical, expectedLogical);
}

@Test void test() {
  // Equivalent SQL:
  //   SELECT DISTINCT *
  //   FROM emp
  final RelBuilder builder = builder();
  final RelNode rel =
      builder.scan("EMP")
          .build();
  String expectedPhysical = ""
      + "EnumerableTableScan(table=[[scott, EMP]])\n";
  String expectedLogical = ""
      + "LogicalTableScan(table=[[scott, EMP]])\n";
  verify(rel, expectedPhysical, expectedLogical);
} 

I think it can be improved to something like

Logical Project (DISTINCT)
+- Logical Scan (Table: employees)

 

 

Issue Links

• relates to: CALCITE-1634 Make RelBuilder.distinct no-op if input is already distinct