Ozan Gunalp

I don't think that swallowing the Kafka client `ConfigException` or reactive messaging deployment exception is a good idea. I guess it would be possible to initialize producer/consumer clients lazily for...

@davebarda ok I've drafted something, will reference you to a PR later today

On the consumer side, we already ignore the health check if there are no subscribers. But on the producer side indeed there would be a change of behaviour. I was...

@gsmet I finally re-wrote the substitution for the class `GenericDatumReader` from Avro to match the latest version. But If we remove the substitution altogether the native compiler no longer complains...

That's odd, I didn't expected that infinispan failure was related. Without the substitution the avro related native compilations passed. I'll push the fixed substitution rather than removing it.

@lennartj the original issue you created and the Emitter Factory feature are two different things. Reactive Messaging works on CDI-managed beans and detects declared channels/emitters in order to create necessary...

I would add to the previous list: - Maybe it goes without saying but having a container util for running the nats server in tests is very important. - Backpressure...

The approach implemented in #1680 uses the emitter pattern for writing Kafka records inside transactions. It defines a custom emitter called `KafkaTransactions`, which can be injected as you'd do for...

@cdennison Thanks for your PR. I think you need to add `merge=true` to the `to-rabbitmq-string` channel for the Emitter to channel to work.

@cdennison the quickstart code seems correct. I had an issue with the rabbbitmq image you used: the connections were established but no message transmission was taking place. I think there...