Bevy version
0.14.0
What you did
use bevy::prelude::*;
fn main() {
App::new()
.add_plugins(MinimalPlugins)
.add_systems(Startup, startup)
.add_systems(Update, update)
.run();
}
#[derive(Component)]
struct Marker;
fn startup(mut commands: Commands) {
commands.spawn(Marker).observe(|_: Trigger<OnRemove, Marker>, mut commands: Commands| {
commands.spawn_empty();
});
}
fn update(
query: Query<Entity, With<Marker>>,
mut commands: Commands
) {
for entity in &query {
commands.entity(entity).despawn();
}
}What went wrong
What were you expecting?
- Entity with
Marker component gets queued for despawning.
- Its observer runs and queues the spawning of a new entity.
- End result is the entity with
Marker being despawned, and a new empty entity being spawned.
- App continues to run.
What actually happened?
thread 'main' panicked at F:\packages\cargo\registry\src\index.crates.io-6f17d22bba15001f\bevy_ecs-0.14.0\src\entity\mod.rs:582:9:
flush() needs to be called before this operation is legal
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
Encountered a panic when applying buffers for system `flush_panic_repro::update`!
Encountered a panic in system `bevy_app::main_schedule::Main::run_main`!
error: process didn't exit successfully: `target\debug\flush_panic_repro.exe` (exit code: 101)
Additional information
#14300 runs into the same panic, but involves component hooks, MapEntities, and DynamicScene. I don't know enough about Bevy's internals to know if these are ultimately the same bug, or just coincidentally panic in the same spot.
If just the Marker component is removed without despawning its entity, the observer still runs (as expected) but there is no panic:
fn update(
query: Query<Entity, With<Marker>>,
mut commands: Commands
) {
for entity in &query {
commands.entity(entity).remove::<Marker>(); // Does NOT panic.
}
}
// Rest of code is unchanged.
Here's what seems like the most relevant part of the panic backtrace:
2: bevy_ecs::entity::Entities::verify_flushed
at F:\packages\cargo\registry\src\index.crates.io-6f17d22bba15001f\bevy_ecs-0.14.0\src\entity\mod.rs:582
3: bevy_ecs::entity::Entities::free
at F:\packages\cargo\registry\src\index.crates.io-6f17d22bba15001f\bevy_ecs-0.14.0\src\entity\mod.rs:680
4: bevy_ecs::world::entity_ref::EntityWorldMut::despawn
at F:\packages\cargo\registry\src\index.crates.io-6f17d22bba15001f\bevy_ecs-0.14.0\src\world\entity_ref.rs:1235
5: bevy_ecs::world::World::despawn
at F:\packages\cargo\registry\src\index.crates.io-6f17d22bba15001f\bevy_ecs-0.14.0\src\world\mod.rs:1105
6: bevy_ecs::system::commands::despawn
at F:\packages\cargo\registry\src\index.crates.io-6f17d22bba15001f\bevy_ecs-0.14.0\src\system\commands\mod.rs:1241
Bevy version
0.14.0What you did
What went wrong
What were you expecting?
Markercomponent gets queued for despawning.Markerbeing despawned, and a new empty entity being spawned.What actually happened?
Additional information
#14300 runs into the same panic, but involves component hooks,
MapEntities, andDynamicScene. I don't know enough about Bevy's internals to know if these are ultimately the same bug, or just coincidentally panic in the same spot.If just the
Markercomponent is removed without despawning its entity, the observer still runs (as expected) but there is no panic:Here's what seems like the most relevant part of the panic backtrace: