pack-priv final Class UniqueWithDuplicateNullsMergeSort

Class Inheritance

java.lang.Object
org.apache.derby.impl.store.access.sort.MergeSort
org.apache.derby.impl.store.access.sort.UniqueWithDuplicateNullsMergeSort

Imports

org.apache.derby.shared.common.error.StandardException, org.apache.derby.iapi.types.DataValueDescriptor

This class extends and customizes MergeSort to support unique indexes with duplicate nulls. It overrides compare method to consider keypart - 1 parts of the keys while comparing (only for non null keys).

Field Summary

Inherited from org.apache.derby.impl.store.access.sort.MergeSort:: alreadyInOrder columnOrdering columnOrderingAscendingMap columnOrderingMap columnOrderingNullsLowMap properties sortBufferMax sortBufferMin sortObserver template

Constructor Summary

Access	Constructor and Description
pack-priv	UniqueWithDuplicateNullsMergeSort()

Method Summary

Modifier and Type	Method and Description
protected int	Returns: 0 for duplicates non zero for distinct keys compare(DataValueDescriptor[] keys r1, DataValueDescriptor[] keys r2) Overrides org.apache.derby.impl.store.access.sort.MergeSort.compare. Compares two keys.

Inherited from org.apache.derby.impl.store.access.sort.MergeSort:: checkColumnTypes createMergeRun doneInserting doneScanning doneScanning drop dropMergeRuns initialize open openSortRowSource openSortScan

Constructor Detail

UniqueWithDuplicateNullsMergeSort	back to summary
pack-priv UniqueWithDuplicateNullsMergeSort()

Method Detail

compare	back to summary
protected int compare(DataValueDescriptor[] r1, DataValueDescriptor[] r2) throws StandardException Overrides org.apache.derby.impl.store.access.sort.MergeSort.compare. Compares two keys. If all the parts of the keys are not null then the leading (keys.length - 1) parts are compared, else if a part of the key is null then all parts of the key are compared (keys.length). This behavior is useful for implementing unique constraints where multiple null values are allowed, but uniqueness must still be guaranteed for keys with no null values. In this case the leading parts of the key are the user key columns, while the last column is a system provided column which is guaranteed unique per base row. Parameters r1:DataValueDescriptor[] keys r2:DataValueDescriptor[] keys Returns:int 0 for duplicates non zero for distinct keys Annotations @Override

compare

back to summary

protected int compare(DataValueDescriptor[] r1, DataValueDescriptor[] r2) throws StandardException

Overrides org.apache.derby.impl.store.access.sort.MergeSort.compare.

Compares two keys. If all the parts of the keys are not null then the leading (keys.length - 1) parts are compared, else if a part of the key is null then all parts of the key are compared (keys.length). This behavior is useful for implementing unique constraints where multiple null values are allowed, but uniqueness must still be guaranteed for keys with no null values. In this case the leading parts of the key are the user key columns, while the last column is a system provided column which is guaranteed unique per base row.

Parameters

r1:DataValueDescriptor[]: keys
r2:DataValueDescriptor[]: keys

Returns:int

0 for duplicates non zero for distinct keys

Annotations

@Override