Wrong example usage of CompletableFututure's async methods

[Bastian]

We have examples like this in the docs:

server.createTextChannelBuilder()
    .setName("new-channel")
    .create()
    .thenAcceptAsync(channel -> {
        // This code is already running async, so you can safely use join() here
        Message message = channel.sendMessage("First!").join();
        message.addReaction("👍").join();
    });

thenAcceptAsync(...) and other async methods use the ForkJoinPool.commonPool() by default. This pool is bound to the nuber of cpu cores and is not meant to be used for blocking operations.

[Vampire]

While I second using proper asynchronous programming instead of blocking on additional threads, what is it meant to be used for if not for long-running or blocking operations?
As far as I understood it, the ...Async methods are exactly for not blocking the thread the completes the future if the action is long-running or blocking, using the asynchronous strategy of the completable future which by default is the common fork join pool, yes.

[Bastian]

It's meant for computational tasks, not for blocking tasks. E.g., parallel streams is a perfect example that uses the common pool, too.

[Vampire]

What would you suggest for a blocking task instead?
I considered blocking and long running (aka computational) the same in this regard.
You let the CF do it on another thread to not block the calling thread.

[Bastian]

What would you suggest for a blocking task instead?

Either use an own or Javacord's executor service or just don't block.

I considered blocking and long running (aka computational) the same in this regard.

The common pool size is limited to the core count of the machine, so there's a difference.

See https://stackoverflow.com/a/46073118

[Vampire]

Ah, yes, makes sense of course.