Subject: Using Spark Accumulators with Structured Streaming

Yes it is application specific class. This is how java Spark Functions work.
You can refer to this code in the documentation:

public class StateUpdateTask implements MapGroupsWithStateFunction<String,
InputEventModel, ModelStateInfo, ModelUpdate> {

    public ModelUpdate call(String productId, Iterator<InputEventModel>
eventsIterator, GroupState<ModelStateInfo> state) {

On Thu, May 28, 2020 at 10:59 PM Something Something <