comnetwork commented on a change in pull request #444: PHOENIX-5148 URL: https://github.com/apache/phoenix/pull/444#discussion_r265406629
########## File path: phoenix-core/src/main/java/org/apache/phoenix/compile/OrderPreservingTracker.java ########## @@ -9,29 +9,37 @@ */ package org.apache.phoenix.compile; +import java.sql.SQLException; import java.util.ArrayList; import java.util.Collections; import java.util.Comparator; +import java.util.HashMap; import java.util.Iterator; +import java.util.LinkedList; import java.util.List; +import java.util.Map; +import org.apache.hadoop.hbase.util.Pair; import org.apache.phoenix.compile.GroupByCompiler.GroupBy; -import org.apache.phoenix.execute.TupleProjector; +import org.apache.phoenix.compile.OrderByCompiler.OrderBy; import org.apache.phoenix.expression.CoerceExpression; import org.apache.phoenix.expression.Expression; import org.apache.phoenix.expression.KeyValueColumnExpression; import org.apache.phoenix.expression.LiteralExpression; +import org.apache.phoenix.expression.OrderByExpression; import org.apache.phoenix.expression.ProjectedColumnExpression; import org.apache.phoenix.expression.RowKeyColumnExpression; import org.apache.phoenix.expression.RowValueConstructorExpression; import org.apache.phoenix.expression.function.FunctionExpression.OrderPreserving; import org.apache.phoenix.expression.function.ScalarFunction; import org.apache.phoenix.expression.visitor.StatelessTraverseAllExpressionVisitor; import org.apache.phoenix.expression.visitor.StatelessTraverseNoExpressionVisitor; +import org.apache.phoenix.jdbc.PhoenixConnection; import org.apache.phoenix.schema.PTable; -import org.apache.phoenix.schema.SortOrder; +import org.apache.phoenix.schema.TableRef; import org.apache.phoenix.util.ExpressionUtil; +import com.google.common.collect.ImmutableList; import com.google.common.collect.Iterators; import com.google.common.collect.Lists; Review comment: There are mainly three changes for the class `OrderPreservingTracker`: 1. I added a `getInputOrderBys` method to determine the input `OrderBys` by the combination of `innerQueryPlan`'s output `OrderBys` ， `GroupBy` of current QueryPlan and the `rowKeyColumns` of current table. 2. Because the innerQueryPlan may have multiple output OrderBys(for `SortMergeJoinPlan`), so I extracted many stateful memeber variables in `OrderPreservingTracker`(such as `orderPreservingTrackInfos`, `isOrderPreserving` and `isReverse` etc.) to a new inner class `TrackOrderByContext`, `TrackOrderByContext` is used to track if a single Input OrderBy matchs the target OrderByExpressions in `OrderPreservingTracker.isOrderPreserving()` method, and once we found a `TrackOrderByContext` satisfied `isOrderPreserving()`, `OrderPreservingTracker.selectedTrackOrderByContext` member variable is set to this `TrackOrderByContext`, then we can use this `selectedTrackOrderByContext` to implement the `OrderPreservingTracker.getOrderPreservingTrackInfos` and `OrderPreservingTracker. isReverse` methods etc. BTW. at most only one `TrackOrderByContext` can meet `isOrderPreserving` is true. 3. I added `ascending` and `nullsLast` to the inner class` Info` , and extracted the complete ordering information in `Info` class by the inner `TrackOrderPreservingExpressionVisitor` class, so we can inferring alignment between the target OrderByExpressions and the input OrderBys based on `Info` in `TrackOrderByContext.doTrack` method， not on the row keys like the original `OrderPreservingTracker.track` method does. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
