org.apache.flink.statefun.flink.state.processor.union

Class BootstrapDatasetUnion

java.lang.Object

org.apache.flink.statefun.flink.state.processor.union.BootstrapDatasetUnion

public final class BootstrapDatasetUnion
extends java.lang.Object

Utility to union multiple BootstrapDatasets to a single, tagged Flink dataset.

Constructor Summary

Constructors

Constructor and Description
BootstrapDatasetUnion()

Method Summary

Modifier and Type	Method and Description
static org.apache.flink.api.java.DataSet<TaggedBootstrapData>	apply(java.util.List<BootstrapDataset<?>> bootstrapDatasets) Unions multiple BootstrapDatasets to a single, tagged Flink dataset.

Methods inherited from class java.lang.Object

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

Constructor Detail

BootstrapDatasetUnion

public BootstrapDatasetUnion()

Method Detail

apply

public static org.apache.flink.api.java.DataSet<TaggedBootstrapData> apply(java.util.List<BootstrapDataset<?>> bootstrapDatasets)

Unions multiple BootstrapDatasets to a single, tagged Flink dataset.

This does a few things:

• For each BootstrapDataset, assign a union index according to the order that they were given.
• For each BootstrapDataset, tag each element with their assigned union index and target state bootstrap function address (as designated by routers). This is a flat map operation that transforms the original user-registered dataset to a dataset of TaggedBootstrapData.
• Creates a type information that unions all type information of the data types of all user-registered datasets. The union serializer created by this type information multiplex multiple data types by pre-fixing each written record with their corresponding union index.

Parameters:

bootstrapDatasets - pre-tagged, user-registered bootstrap datasets to union.

Returns:

the union, tagged bootstrap dataset.