chore(core): deprecated AsyncApiDocket bean

[timonback]

Use spring properties instead

[netlify]

✅ Deploy Preview for springwolf-ui canceled.

Name	Link
🔨 Latest commit	159c97e
🔍 Latest deploy log	https://app.netlify.com/sites/springwolf-ui/deploys/653bc2a94e4fbb00084e1fa2

[maksymgendin]

@timonback Couldn't find a better place for commenting on this.

We're currently using manual configuration via AsyncApiDocket and I'm not sure what should be an alternative for our use case because we don't use the default Spring Kafka integration where method signature and annotations are used to configure the Kafka consumers, producers and payloads.

We have a fully reactive application with io.projectreactor.kafka:reactor-kafka where the entry point in consuming Kafka records is a ReactiveKafkaConsumerTemplate:

private <T> ReactiveKafkaConsumerTemplate<String, T> buildReactiveKafkaConsumerTemplate(Class<T> eventClass,
                                                                                        KafkaProperties kafkaProperties,
                                                                                        String topicName,
                                                                                        MeterRegistry meterRegistry) {
    return new ReactiveKafkaConsumerTemplate<>(
            ReceiverOptions.<String, T>create(kafkaProperties.buildConsumerProperties())
                    .withKeyDeserializer(new ErrorHandlingDeserializer<>(new StringDeserializer()))
                    .withValueDeserializer(new ErrorHandlingDeserializer<>(new JsonDeserializer<>(eventClass, false)))
                    .consumerListener(new MicrometerConsumerListener(meterRegistry))
                    .subscription(Collections.singleton(topicName))
    );
}

ReceiverOptions is a class from reactor-kafka as well.

Could you please give me any hints how Springwolf is supposed to work with configuration only in our case?

Update: I found a working solution for us:

@Configuration
@Profile("local")
class LocalSpringwolfConfiguration {

    @AsyncListener(
            operation = @AsyncOperation(
                    channelName = "${kafka.topics.my-topic.topic-name}",
                    payloadType = MyEvent.class
            )
    )
    @KafkaAsyncOperationBinding
    void myTopicConfiguration() {

    }
}

and

springwolf:
  enabled: true
  use-fqn: true # has to be set, otherwise an HTTP 400 is thrown when publishing from browser
  docket:
    base-package: local
    info:
      title: ${spring.application.name}
      version: dummy
    servers:
      local:
        url: ${spring.kafka.bootstrap-servers}
        protocol: kafka
  plugin:
    kafka:
      publishing:
        enabled: true
        producer:
          client-id: ${spring.application.name}
          bootstrap-servers: ${spring.kafka.bootstrap-servers}
          key-serializer: org.apache.kafka.common.serialization.StringSerializer
          value-serializer: org.springframework.kafka.support.serializer.JsonSerializer

[maksymgendin]

One more problem is that we're using Springwolf mainly for local development with Testcontainers and the servers property is currently set dynamically on startup when constructing the AsyncApiDocket bean. How this can work with Spring properties?

Update: This problem is actually solvable like this:

springwolf:
  servers:
    local:
      url: ${spring.kafka.bootstrap-servers}
      protocol: kafka

[timonback]

Hi @maksymgendin,
feel free to comment in #445 in case you run into issue. We improved our migration guide.