Flux::flatMap何时并发监听多数据源？触发场景解析

Great question! Based on hands-on experimentation, here's the specific scenario that triggers this behavior, along with key details:

The Critical Scenario for Concurrent Listening #

Flux::flatMap will end up concurrently listening to all its internal streams (ranging from 0 up to an unlimited number) when the upstream thread sending signals to flatMap is completely separate from every thread used by the internal streams.

To make this concrete with an example:

• Imagine your upstream source runs on a thread named thread-upstream-1, pushing events into flatMap.
• You have N internal streams, each emitting their own signals on unique threads: thread-inner-stream-1, thread-inner-stream-2, ..., thread-inner-stream-N (where 1≤i≤N).
• If thread-upstream-1 is not identical to any of the thread-inner-stream-i threads, flatMap will automatically handle all these internal streams concurrently.

Quick Context on Why This Works #

flatMap relies on thread context to manage concurrency boundaries. When the upstream thread doesn't overlap with any internal stream threads, there's no thread affinity constraint forcing serialized processing. This lets flatMap spin up concurrent handlers for each internal stream, leading to the 0-to-unlimited concurrent listening behavior observed.

内容的提问来源于stack exchange，提问作者Stav Alfi