fix: handle SIGTERM

[mxfactorial]

closes #220 with https://github.com/tokio-rs/axum/blob/6d78e843982ca299b5d15b7b9095c249e6cdf978/examples/graceful-shutdown/src/main.rs#L56C1-L76C6

[4t145]

We shouldn't handle specific signal here, we should wait for those signal outside and cancel the cancellation token.

[mxfactorial]

here vs "outside" is another issue

this pr lets the process owner shut down their axum server

[4t145]

let me explain this, if you want to use graceful shutdown, you should await the shutdown signal and then trigger the cancellation token.

{
    let ct = CancellationToken::new();
    let config = new_config_with_ct(ct.clone());
    tokio::spawn(async move {
            graceful_shutdown().await;
            ct.cancel();
    });
    // create server
}

In your code, you firstly await cancellation token, and then waiting for shutdown signal, but ct has already been in responsibility of shutdown signal.

[mxfactorial]

merge the code so process owners can shut down this axum server

"but they need to use CancellationToken"

no they don't

[4t145]

Can you provide some concrete example of why you need to put this code here, personally I don't see any necessity.

The disign here is user use cancellation token to shutdown the server. You cannot forcing people wait ctrl_c or SIGTERM, it is not always the case.

[mxfactorial]

SIGTERM and cntrl c are concrete enough examples

and the code gives owners simple control over their processes

they dont need disign

thats your last response

[4t145]

Sorry maybe I'm not smart enough to understand you. Since you said that's your last response, I will close this pr.