Package org.apache.drill.exec.physical.impl.xsort

Class MSortTemplate

java.lang.Object

org.apache.drill.exec.physical.impl.xsort.MSortTemplate

All Implemented Interfaces:

MSorter, org.apache.hadoop.util.IndexedSortable

public abstract class MSortTemplate
extends Object
implements MSorter, org.apache.hadoop.util.IndexedSortable

Field Summary

Fields inherited from interface org.apache.drill.exec.physical.impl.xsort.MSorter

TEMPLATE_DEFINITION

Constructor Summary

Constructors

Constructor	Description
MSortTemplate()

Method Summary

Modifier and Type	Method	Description
void	clear()
int	compare(int leftIndex, int rightIndex)
abstract int	doEval(int leftIndex, int rightIndex)
abstract void	doSetup(FragmentContext context, VectorContainer incoming, RecordBatch outgoing)
SelectionVector4	getSV4()
static long	memoryNeeded(int recordCount)	For given recordCount how much memory does MSorter needs for its own purpose.
protected int	merge(int leftStart, int rightStart, int rightEnd, int outStart)	Given two regions within the selection vector 4 (a left and a right), merge the two regions to produce a combined output region in the auxiliary selection vector.
void	setup(FragmentContext context, BufferAllocator allocator, SelectionVector4 vector4, VectorContainer hyperBatch, int outputBatchSize, int desiredBatchSize)
void	sort()	Merge a set of pre-sorted runs to produce a combined result set.
void	swap(int sv0, int sv1)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

MSortTemplate

public MSortTemplate()

Method Details

setup

public void setup(FragmentContext context,
 BufferAllocator allocator,
 SelectionVector4 vector4,
 VectorContainer hyperBatch,
 int outputBatchSize,
 int desiredBatchSize)
           throws SchemaChangeException

Specified by:

setup in interface MSorter

Throws:

SchemaChangeException

memoryNeeded

public static long memoryNeeded(int recordCount)

For given recordCount how much memory does MSorter needs for its own purpose. This is used in ExternalSortBatch to make decisions about whether to spill or not.

merge

protected int merge(int leftStart,
 int rightStart,
 int rightEnd,
 int outStart)

Given two regions within the selection vector 4 (a left and a right), merge the two regions to produce a combined output region in the auxiliary selection vector.

getSV4

public SelectionVector4 getSV4()

Specified by:

getSV4 in interface MSorter

sort

public void sort()

Merge a set of pre-sorted runs to produce a combined result set. Merging is done in the selection vector, record data does not move.

Runs are merge pairwise in multiple passes, providing performance of O(n * m * log(n)), where n = number of runs, m = number of records per run.

Specified by:

sort in interface MSorter

swap

public void swap(int sv0,
 int sv1)

Specified by:

swap in interface org.apache.hadoop.util.IndexedSortable

compare

public int compare(int leftIndex,
 int rightIndex)

Specified by:

compare in interface org.apache.hadoop.util.IndexedSortable

clear

public void clear()

Specified by:

clear in interface MSorter

doSetup

public abstract void doSetup(@Named("context")
 FragmentContext context,
 @Named("incoming")
 VectorContainer incoming,
 @Named("outgoing")
 RecordBatch outgoing)
                      throws SchemaChangeException

Throws:

SchemaChangeException

doEval

public abstract int doEval(@Named("leftIndex")
 int leftIndex,
 @Named("rightIndex")
 int rightIndex)
                    throws SchemaChangeException

Throws:

SchemaChangeException