#[traced_test] fails with error

1

I've previously used #[traced_test] with no issues, but, in a current project using [tokio_test] along with [serial], tests fail with this error:

Could not set global tracing subscriber: SetGlobalDefaultError("a global default trace dispatcher has already been set")

No test anywhere in the project is setting any global default trace dispatcher.

I can't find a bit about this on the web, and I have been unable to track it down to a minimal reproducible case.

Here's the odder part:

  • If I run a single test without --nocapture, the test works (no above error) - but there is no tracing output
  • If I run a single test with --nocapture, the error occurs
  • If I run all tests, it happens, with or without --nocapture
2

Are you talking about combining serial-test with tracing-test? I just tried it using the latest versions of all crates (and tokio's full feature), and it's working in a minimal demo (with and without --nocapture):

fn main() {
    println!("Hello, world!");
}

#[cfg(test)]
mod tests {
    use serial_test::serial;
    use tracing::info;
    use tracing_test::traced_test;

    #[traced_test]
    #[tokio::test]
    #[serial]
    async fn test_one() {
        tokio::time::sleep(std::time::Duration::from_secs(1)).await;
        info!("one");
    }

    #[traced_test]
    #[tokio::test]
    #[serial]
    async fn test_two() {
        tokio::time::sleep(std::time::Duration::from_secs(1)).await;
        info!("two");
    }
}

As expected, each test pauses for 1 second before printing ok. And if I make either panic, I see the tracing info log.

Your best bet is going to be coming up with a minimal repro. treereduce-rust might be able to help automate some of that.

3

As I said, tokio-test is in the mix too.
Non-async tests work fine.
And I've had it work in the past, too.
It no longer works.
I am certain there is an issue. I was hoping someone else had run into it and had some ideas.

4

I'm also using tokio::test in the code above, and there is no problem there.

I'm trying to say that I am unable to reproduce your results, and you might have to do some work to minimize a reproduction. Especially since you haven't come across anyone else with similar reports.

5

I now know that's the case. :slight_smile:
Yes, I'll have to come up with a MRS. At some point.
For now, I think I'll create a new set of macros that wraps the tracing ones and adds println!.