Inconsistent Server Stop Logic

[Yimura]

I've searched existing issues and couldn't find a duplicate.

• I confirm this is not a duplicate.

Operating System

Any

Server Software Version/Commit

master

What happened?

The stop command has a fairly normal logic of stopping the server and handling everything:

Pumpkin/pumpkin/src/command/commands/cmd_stop.rs

Lines 31 to 36 in febbfd8

	let kick_message = TextComponent::text("Server stopped");
	for player in server.get_all_players().await {
	player.kick(kick_message.clone()).await;
	}
	server.save().await;
	std::process::exit(0)

However the signal handler is missing these player kicking and saving steps:

Pumpkin/pumpkin/src/main.rs

Lines 258 to 266 in febbfd8

	fn handle_interrupt() {
	log::warn!(
	"{}",
	TextComponent::text("Received interrupt signal; stopping server...")
	.color_named(NamedColor::Red)
	.to_pretty_console()
	);
	std::process::exit(0);
	}

To Reproduce

Using the stop command results in a panic in the world saving logic (most likely because this hasn't been implemented but that's beside the point of this issue).

While ^C/SIGINT does not cause this crash and closes the server properly.

I'm mainly trying to highlight the difference in behaviour between the two shutdown methods.

Expected behavior

Centralise the logic of shutting the server down so there's no discrepancies between the server handling a stop command and it receiving a ^C/SIGINT/SIGTERM from the command line.

[Polygons1]

stop issues are caused because of the dashmap crate, I have debugged it.

[Snowiiii]

I put that on the to do list already, I think we should use channels for something like that to send shutdowns when we need them. Also https://tokio.rs/tokio/topics/shutdown is useful

[Snowiiii]

I would be happy if someone may want to make a PR