[Solved] BevyError is currently not a more convenient choice than .unwrap()

[lomirus]

Note

Update: This discussion is no longer valid, because it could be solved by enabling the short backtrace by the solution from @MrGVSV.

---

In 0.16, Bevy has made the Unified ECS Error Handling. It looks fascinating, so I tried the 0.16 and compared the error log between the new and old error handling paradigm:

Old:

fn update(query: Query<&A>) {
    let _ = query.single().unwrap();
}

New:

fn update(query: Query<&A>) -> Result {
    let _ = query.single()?;

    Ok(())
}

I did not spawn any A component, and there is their error log:

Old:

thread 'Compute Task Pool (0)' panicked at src\main.rs:7:28:
called `Result::unwrap()` on an `Err` value: NoEntities("bevy_ecs::system::query::Query<&bevy_demo::A>")
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
Encountered a panic in system `bevy_demo::update`!
Encountered a panic in system `bevy_app::main_schedule::Main::run_main`!
error: process didn't exit successfully: `target\debug\bevy-demo.exe` (exit code: 101)

New:

thread 'Compute Task Pool (5)' panicked at C:\Users\pc\.cargo\registry\src\index.crates.io-1949cf8c6b5b557f\bevy_ecs-0.16.0\src\error\handler.rs:141:1:
Encountered an error in system `bevy_demo::update`: No entities fit the query bevy_ecs::system::query::Query<&bevy_demo::A>

note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
Encountered a panic in system `bevy_demo::update`!
Encountered a panic in system `bevy_app::main_schedule::Main::run_main`!
error: process didn't exit successfully: `target\debug\bevy-demo.exe` (exit code: 101)

In the old error log, we can quickly locate the error in position src\main.rs:7:28:

let _ = query.single().unwrap();

However in the new paradigm, the information we got is only C:\Users\pc\.cargo\registry\src\index.crates.io-1949cf8c6b5b557f\bevy_ecs-0.16.0\src\error\handler.rs:141:1:, it's does not provide any useful information to the user.

In other words, instead, .unwrap() seems more user-friendly for the developers (as least for now).

I'm not sure if this is an issue with my usage. I have enabled the track_location feature, and the result did not make any difference.

[MrGVSV]

Interesting. Even with default features disabled, I'm getting the following:

thread 'main' panicked at crates/bevy_ecs/src/error/handler.rs:141:1:
Encountered an error in system `hello_world::update`: No entities fit the query bevy_ecs::system::query::Query<&hello_world::A>
   5: hello_world::update
             at ./examples/hello_world.rs:21:13 // <-- Where the error originated

// ... backtrace ... //

Encountered a panic in system `hello_world::update`!
Encountered a panic in system `bevy_app::main_schedule::Main::run_main`!

Process finished with exit code 101

[MrGVSV]

Okay I think I know the issue. You have to set RUST_BACKTRACE.

So instead of:

cargo run

Do:

RUST_BACKTRACE=short cargo run

The reason it worked for me is because RustRover defaults to short backtraces.

---

Also, keep in mind that this is just the default panic behavior. You can set custom behaviors globally or even per-system, such as logging or ignoring the errors.

[lomirus]

Thanks for the answer, that makes sense now.