Why use of Scheduler if the processing has been proven NON blocking please?

Hello,

This is my first ticket in this project.

I would like to start with a big thanks. The talk in Kafka submit was very clear, easy to follow.

I am observing an issue for the reactive parallel consumer please:

https://github.com/confluentinc/parallel-consumer/blob/master/parallel-consumer-reactor/src/main/java/io/confluent/parallelconsumer/reactor/ReactorProcessor.java#L104

May I ask why the publication on the Scheduler please?

Let's suppose the processing of the event is NON blocking, not blocking at all (BlockHound tested non blocking etc...)

Should the Scheduler be removed, if the processing of the event itself has been assured NON blocking, and still be able to have the parallel reactive consumer?

Thank you

Start with parallel-consumer-reactor/src/main/java/io/confluent/parallelconsumer/reactor/ReactorProcessor.java at line 104. Investigate why publication uses a Scheduler and compare that behavior with the issue's BlockHound-tested non-blocking processing case. Done means the project has a decided, technically justified answer about whether the Scheduler should be removed or made optional.

Written by the indexing model from the issue text.