Schemas

Kafka messages SHOULD use schemas (AVRO, JSON, Protobuf) registered in the Schema Registry to ensure compatibility between producers/consumers.

If so, always refer to the schema definitions directly in the schema registry instead of duplicating the schema definitions inline. This is to avoid double maintenance.

An example directly taken from reference spec is shown below.

...
channels:
  namespace.source.event.topic-name:
    ...
    publish:
      ...
      message:
        $ref: '#/components/messages/namespace.source.event.topic-name'

components:
  ...
  messages:
    namespace.source.event.topic-name:
      $ref: '#/components/schemas/namespace.source.event.topic-name'
  ...
  schemas:
    namespace.source.event.topic-name:
      description: 'Schema[s] retrieved from Schema Registry'
      schemaFormat: 'application/vnd.apache.avro;version=1.9.0'
      payload:
        $ref: 'https://dev-fdp-playground-schema-registry.api.3stripes.io/subjects/namespace.source.event.topic-name-value/versions/latest/schema'
      bindings:
        kafka:
          key:
            $ref: 'https://dev-fdp-playground-schema-registry.api.3stripes.io/subjects/namespace.source.event.topic-name-key/versions/latest/schema'

Important note The used schema is a very simple one, it is only used to illustrate how to refer to it. In case the message doesn't use schema for the key, bindings field can be omitted from the schema.