[CALCITE-5521] Remove redundant rowtype check in RelSubset#add

• Type: Improvement
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: 1.34.0
• Component/s: core
• Labels: pull-request-available
• Assignee: asdfgh19
• Reporter: asdfgh19
• Created: 09/Feb/23 08:46
• Updated: 14/Mar/23 13:53
• Resolved: 09/Feb/23 13:51

Description

void add(RelNode rel) {
  if (set.rels.contains(rel)) {
    return;
  }

  VolcanoPlanner planner = (VolcanoPlanner) rel.getCluster().getPlanner();
  if (planner.getListener() != null) {
    RelOptListener.RelEquivalenceEvent event =
        new RelOptListener.RelEquivalenceEvent(
            planner,
            rel,
            this,
            true);
    planner.getListener().relEquivalenceFound(event);
  }

  // If this isn't the first rel in the set, it must have compatible
  // row type.
  if (set.rel != null) {
    RelOptUtil.equal("rowtype of new rel", rel.getRowType(),
        "rowtype of set", getRowType(), Litmus.THROW);
  }
  set.addInternal(rel);
  if (false) {
    Set<CorrelationId> variablesSet = RelOptUtil.getVariablesSet(rel);
    Set<CorrelationId> variablesStopped = rel.getVariablesSet();
    Set<CorrelationId> variablesPropagated =
        Util.minus(variablesSet, variablesStopped);
    assert set.variablesPropagated.containsAll(variablesPropagated);
    Set<CorrelationId> variablesUsed = RelOptUtil.getVariablesUsed(rel);
    assert set.variablesUsed.containsAll(variablesUsed);
  }
} 

Double checking relset rowtype and input parameter relNode rowtype may be unnecessary

Issue Links

• links to: GitHub Pull Request 3063
• links to: GitHub Pull Request #3063