Mailing-List: contact dev-help@quickstep.incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@quickstep.incubator.apache.org
From: jianqiao <git@git.apache.org>
To: dev@quickstep.incubator.apache.org
Reply-To: dev@quickstep.incubator.apache.org
References: <git-pr-171-incubator-quickstep@git.apache.org>
In-Reply-To: <git-pr-171-incubator-quickstep@git.apache.org>
Subject: [GitHub] incubator-quickstep pull request #171: QUICKSTEP-68 Reorder intermediate rel...
Content-Type: text/plain
Message-Id: <20170130210816.5A0B1DFBDA@git1-us-west.apache.org>
Date: Mon, 30 Jan 2017 21:08:16 +0000 (UTC)
archived-at: Mon, 30 Jan 2017 21:08:57 -0000

Github user jianqiao commented on a diff in the pull request:

    https://github.com/apache/incubator-quickstep/pull/171#discussion_r98541113
  
    --- Diff: query_optimizer/rules/ReorderColumns.cpp ---
    @@ -0,0 +1,205 @@
    +/**
    + * Licensed to the Apache Software Foundation (ASF) under one
    + * or more contributor license agreements.  See the NOTICE file
    + * distributed with this work for additional information
    + * regarding copyright ownership.  The ASF licenses this file
    + * to you under the Apache License, Version 2.0 (the
    + * "License"); you may not use this file except in compliance
    + * with the License.  You may obtain a copy of the License at
    + *
    + *   http://www.apache.org/licenses/LICENSE-2.0
    + *
    + * Unless required by applicable law or agreed to in writing,
    + * software distributed under the License is distributed on an
    + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
    + * KIND, either express or implied.  See the License for the
    + * specific language governing permissions and limitations
    + * under the License.
    + **/
    +
    +#include "query_optimizer/rules/ReorderColumns.hpp"
    +
    +#include <algorithm>
    +#include <cstddef>
    +#include <limits>
    +#include <unordered_map>
    +#include <vector>
    +
    +#include "query_optimizer/expressions/AttributeReference.hpp"
    +#include "query_optimizer/expressions/ExprId.hpp"
    +#include "query_optimizer/expressions/NamedExpression.hpp"
    +#include "query_optimizer/physical/HashJoin.hpp"
    +#include "query_optimizer/physical/PatternMatcher.hpp"
    +#include "query_optimizer/physical/Physical.hpp"
    +#include "query_optimizer/physical/PhysicalType.hpp"
    +#include "query_optimizer/physical/Selection.hpp"
    +#include "query_optimizer/physical/TableReference.hpp"
    +
    +#include "glog/logging.h"
    +
    +namespace quickstep {
    +namespace optimizer {
    +
    +namespace E = ::quickstep::optimizer::expressions;
    +namespace P = ::quickstep::optimizer::physical;
    +
    +P::PhysicalPtr ReorderColumns::apply(const P::PhysicalPtr &input) {
    +  DCHECK(input->getPhysicalType() == P::PhysicalType::kTopLevelPlan);
    +
    +  return applyInternal(input, true);
    +}
    +
    +P::PhysicalPtr ReorderColumns::applyInternal(const P::PhysicalPtr &input,
    +                                             bool lock_ordering) {
    +  // We have to guarantee that the top level ordering of the columns remain
    +  // unchanged so that the output columns are ordered as specified by the user.
    +  // So here we use the flag "lock_ordering" to skip the first transformable
    +  // node (i.e. the first Selection or HashJoin).
    +  const bool is_not_transformable = !IsTransformable(input);
    +  const bool skip_transform = lock_ordering || is_not_transformable;
    +  lock_ordering = lock_ordering && is_not_transformable;
    +
    +  if (skip_transform) {
    +    std::vector<P::PhysicalPtr> new_children;
    +    for (const P::PhysicalPtr &child : input->children()) {
    +      new_children.emplace_back(applyInternal(child, lock_ordering));
    +    }
    +
    +    if (new_children != input->children()) {
    +      return input->copyWithNewChildren(new_children);
    +    } else {
    +      return input;
    +    }
    +  }
    +
    +  // Collect the maximal chain of transformable nodes.
    +  std::vector<P::PhysicalPtr> nodes;
    +  for (P::PhysicalPtr node = input; IsTransformable(node); node = node->children().front()) {
    +    nodes.emplace_back(node);
    +  }
    +  std::reverse(nodes.begin(), nodes.end());
    +
    +  // A greedy algorithm that reorders the output attributes based on the GEN/KILL
    +  // intervals. This algorithm works well with SSB/TPCH queries and is not likely
    +  // to make the plans worse for whatever queries.
    +  std::unordered_map<E::ExprId, std::size_t> base, gen, kill;
    +
    +  const P::PhysicalPtr base_node =
    +      applyInternal(nodes.front()->children().front(), false);
    +  P::TableReferencePtr base_table;
    +  if (P::SomeTableReference::MatchesWithConditionalCast(base_node, &base_table)) {
    +  }
    --- End diff --
    
    Removed.


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---