batch-processing.yaml

# Copyright (C) 2026 Antony Stubbs and contributors

schema_version: 1
kind: feature
title: Batch processing
category: processing
module: parallel-consumer-core
maturity: production-use
availability:
  status: published
  since: 0.5.0.0
  evidence:
    tag: 0.5.0.0
    path: README.adoc
    basis: The PollContext change made batch support available across the processing APIs.
readme_anchor: batching
summary: Deliver multiple records to one processing invocation when a downstream operation benefits from batching.
use_this_when:
  - Your downstream API accepts batches, such as a bulk indexing or write endpoint.
  - You can handle a batch retry as a unit of work.
setup:
  option: ParallelConsumerOptions.batchSize
  rule: Choose a batch-processing API after setting a batch size.
opt_in:
  mechanism: Set batchSize and use a batch processing API.
  isolated_from_stable_modules: false
boundaries:
  - batchSize must be at least 1, which is also the default and the non-batching configuration; ParallelConsumerOptions.validate() rejects zero, a negative and null with an IllegalArgumentException at construction. Before that bound, zero was accepted and the consumer started cleanly and processed nothing.
  - Single-record PollContext accessors and normal processing APIs are invalid while batching is configured.
  - An exception returns every record in the batch to retry; the next attempt need not contain the same batch grouping.
references:
  - label: Core batch example
    path: parallel-consumer-examples/parallel-consumer-example-core/src/main/java/bz/stub/parallelconsumer/examples/core/CoreApp.java