[CALCITE-5563] Reduce loops to optimize RelSubset#getParents and RelSubset#getParentSubsets

• Type: Improvement
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: 1.35.0
• Component/s: core
• Labels: pull-request-available
• Assignee: asdfgh19
• Reporter: asdfgh19
• Created: 07/Mar/23 13:44
• Updated: 27/Jul/23 10:22
• Resolved: 21/Apr/23 13:39

Description

 

/**
 * Returns the collection of RelNodes one of whose inputs is in this
 * subset.
 */
Set<RelNode> getParents() {
  final Set<RelNode> list = new LinkedHashSet<>();
  for (RelNode parent : set.getParentRels()) {
    for (RelSubset rel : inputSubsets(parent)) {
      // see usage of this method in propagateCostImprovements0()
      if (rel == this) {
        list.add(parent);
        break;
      }
    }
  }
  return list;
}

/**
 * Returns the collection of distinct subsets that contain a RelNode one
 * of whose inputs is in this subset.
 */
Set<RelSubset> getParentSubsets(VolcanoPlanner planner) {
  final Set<RelSubset> list = new LinkedHashSet<>();
  for (RelNode parent : set.getParentRels()) {
    for (RelSubset rel : inputSubsets(parent)) {
      if (rel.set == set && rel.getTraitSet().equals(traitSet)) {
        list.add(planner.getSubsetNonNull(parent));
        break;
      }
    }
  }
  return list;
}

 

Once we have found a matching Relsubset from its parent input, the subsequent inner loop is unnecessary，so we can immediately end the inner loop to save a little time. This is just a minor improment.

Issue Links

• links to: GitHub Pull Request #3098