`boa_cli` 100% single-core CPU usage while idling

[zhuzhu81998]

When i say idle, i mean cargo run and do nothing.

Debug build: (no command except ctrl+D before waiting for about 5s)

Release-dbg build:

At least on my machine (linux), it seems that something is getting polled constantly, such that single-core cpu usage is 100%

[zhuzhu81998]

Yeah the loop here is getting bombed:

boa/cli/src/executor.rs

Lines 115 to 117 in e4bbd39

	if let Some(Err(e)) = future::poll_once(group.next()).await.flatten() {
	self.printer.print(uncaught_job_error(&e));
	}

[jedel1043]

Should be fixed by #4994

[linisha15]

Hi! @zhuzhu81998 , I reproduced the issue by:

Building the CLI with cargo build -p boa_cli
Running the REPL with cargo run -p boa_cli
Leaving it idle without executing any code
Opening Task Manager and monitoring the CPU % for the boa.exe process
Observed Behavior:
The idle REPL consumed approximately 0-1% CPU, The event loop appears to be yielding control correctly to the OS instead of busy-spinning.

Picture attached below showing Task Manager CPU monitoring during idle mode.

The event loop looks like it's working fine on my end, but wanted to let you know what I found.
I use windows.

[zhuzhu81998]

gonna have a look later. currently struggling with leetcode 😑

[jedel1043]

@linisha15 0.3% usage is still a substantial amount for a program that's not doing anything, so the bug is still there, it's just not that extreme

BTW #4994 makes it such that the program consistently uses 0.0% CPU when idle

[linisha15]

@jedel1043 thanks , i will take a look.

[zhuzhu81998]

0.3% usage is still a substantial amount for a program that's not doing anything, so the bug is still there, it's just not that extreme

and what i meant was that it consumes 100% of one core. afaik task manager in windows reports total usage here.

And #4994 does seem to make it go to 0% 👍

(now its back to leetcode...)