Files
Alice Cecile b265dc042a Merge SystemParam::validate_param into SystemParam::get_param (#23225)
# Objective

As raised in
[#23174](https://github.com/bevyengine/bevy/pull/23174#discussion_r2868030355_),
we currently duplicate working when looking up our system parameters:
once during validation, and then again when actually fetching the data.

This is (maybe) slow, and would worsen the performance regression
incurred by resources-as-components (#19731).

This strategy also imposes some non-trivial complexity and
maintainability costs. Because "validate" is a distinct step from "use",
it's possible to skip validation! As far as I could tell, this is the
case in a number of places before this PR: particularly in the
unconventional "please just run my system" path. While in most cases
this will simply result in a crash in a different place, it causes these
paths to not handle

Fixes #23179. Fixes #15505.

## Solution

Fundamentally, what we're doing is rolling the
`SystemParam::validate_param` behavior into `SystemParam::get_param`, by
making the latter return a `Result`.

However, there is a tremendous amount of splash damage required to get
that to actually compile and expose the correct semantics. The most
important of these are:

- `SystemState::get` and friends now returns a `Result`
- this leads to a fair bit of assorted unwrap spam in our tests and
weird internal usages
- these tests can probably be refactored to not use `SystemState`
directly in the future now that we have better tools like
`run_system_once`, but eh, not this PR's job
- this is semantically correct, as these params could fail validation
- `System::validate_param_unsafe` has been removed, and validation now
occurs inside of `System::run_unsafe`
- very much a net positive for both abstract robustness and current
correctness
- this impacts the strategy that various executors use: see the next
section

There are a *lot* of moving parts here: I'm sorry that I couldn't get
this into a smaller, more incremental refactor. When reviewing this PR,
you should begin with the migration guide to help get you oriented on
the details: `validation_merging.md`.

From there, the most important files to review are:

1. `system_param.rs`: trait changes and implementers
2. `function_system.rs`: primary implementer of `System`
3. `multithreaded.rs`: the parallel executor

**NOTE TO REVIEWERS:** Please make comments to generate threads; this PR
review might get fairly hairy.

### Performance discussion

For the parallel `MultithreadedExecutor`, validation was previously done
as a cheap pre-validation step,
while checking run conditions.
Now, tasks will be spawned for systems which would fail or are skipped
during validation.

In most cases, avoiding the extra overhead of looking up the required
data twice should dominate.
However, this change may negatively affect systems which are frequently
skipped (e.g. due to `Single`).

### Paths not taken

In this PR, I've decided not to:

- Add another variant
[RunSystemError](https://docs.rs/bevy/latest/bevy/ecs/system/enum.RunSystemError.html),
distinguishing "validation failed" from "system ran but returned an
error".
- While reusing
[RunSystemError::Failed](https://docs.rs/bevy/latest/bevy/ecs/system/enum.RunSystemError.html#variant.Failed)
for both cases is messy, this PR is already a bit of a nightmare to
review.
- Return a result from `ParamSet::get_mut`.
   - Instead, we just `unwrap`.
- Bubbling up the `Result` is technically more correct, but these were
already panicking before if e.g. a resource is missing, and `ParamSet`
is already an ergonomic abomination.

## Testing

I've added a number of new tests to exercise the system param validation
paths, ensuring that validation is done when systems are run.

However, I would appreciate some help benchmarking the net impact on
realistic-ish scenes. `breakout`, `bevy_city` and `many_animated_foxes`
are probably a decent scattering, but I'd be very open to other
suggestions.

Having done this refactor, I think that it's a net improvement for
robustness and clarity even without the perf benefits however, and that
we should proceed unless this is a clear regression.

---------

Co-authored-by: Chris Russell <8494645+chescock@users.noreply.github.com>
Co-authored-by: Kevin Chen <chen.kevin.f@gmail.com>
2026-03-11 19:03:10 +00:00

162 lines
6.7 KiB
Rust

//! This example demonstrates how to use Bevy's ECS and the [`AsyncComputeTaskPool`]
//! to offload computationally intensive tasks to a background thread pool, process them
//! asynchronously, and apply the results across systems and ticks.
//!
//! Unlike the channel-based approach (where tasks send results directly via a communication
//! channel), this example uses the `AsyncComputeTaskPool` to run tasks in the background,
//! check for their completion, and handle results when the task is ready. This method allows
//! tasks to be processed in parallel without blocking the main thread, but requires periodically
//! checking the status of each task.
//!
//! The channel-based approach, on the other hand, detaches tasks and communicates results
//! through a channel, avoiding the need to check task statuses manually.
use bevy::{
ecs::{system::SystemState, world::CommandQueue},
prelude::*,
tasks::{futures::check_ready, AsyncComputeTaskPool, Task},
};
use futures_timer::Delay;
use rand::RngExt;
use std::time::Duration;
fn main() {
App::new()
.add_plugins(DefaultPlugins)
.add_systems(Startup, (setup_env, add_assets, spawn_tasks))
.add_systems(Update, handle_tasks)
.run();
}
// Number of cubes to spawn across the x, y, and z axis
const NUM_CUBES: u32 = 6;
#[derive(Resource, Deref)]
struct BoxMeshHandle(Handle<Mesh>);
#[derive(Resource, Deref)]
struct BoxMaterialHandle(Handle<StandardMaterial>);
/// Startup system which runs only once and generates our Box Mesh
/// and Box Material assets, adds them to their respective Asset
/// Resources, and stores their handles as resources so we can access
/// them later when we're ready to render our Boxes
fn add_assets(
mut commands: Commands,
mut meshes: ResMut<Assets<Mesh>>,
mut materials: ResMut<Assets<StandardMaterial>>,
) {
let box_mesh_handle = meshes.add(Cuboid::new(0.25, 0.25, 0.25));
commands.insert_resource(BoxMeshHandle(box_mesh_handle));
let box_material_handle = materials.add(Color::srgb(1.0, 0.2, 0.3));
commands.insert_resource(BoxMaterialHandle(box_material_handle));
}
#[derive(Component)]
struct ComputeTransform(Task<CommandQueue>);
/// This system generates tasks simulating computationally intensive
/// work that potentially spans multiple frames/ticks. A separate
/// system, [`handle_tasks`], will track the spawned tasks on subsequent
/// frames/ticks, and use the results to spawn cubes.
///
/// The task is offloaded to the `AsyncComputeTaskPool`, allowing heavy computation
/// to be handled asynchronously, without blocking the main game thread.
fn spawn_tasks(mut commands: Commands) {
let thread_pool = AsyncComputeTaskPool::get();
for x in 0..NUM_CUBES {
for y in 0..NUM_CUBES {
for z in 0..NUM_CUBES {
// Spawn new task on the AsyncComputeTaskPool; the task will be
// executed in the background, and the Task future returned by
// spawn() can be used to poll for the result
let entity = commands.spawn_empty().id();
let task = thread_pool.spawn(async move {
let duration = Duration::from_secs_f32(rand::rng().random_range(0.05..5.0));
// Pretend this is a time-intensive function. :)
Delay::new(duration).await;
// Such hard work, all done!
let transform = Transform::from_xyz(x as f32, y as f32, z as f32);
let mut command_queue = CommandQueue::default();
// we use a raw command queue to pass a FnOnce(&mut World) back to be
// applied in a deferred manner.
command_queue.push(move |world: &mut World| {
let (box_mesh_handle, box_material_handle) = {
let mut system_state = SystemState::<(
Res<BoxMeshHandle>,
Res<BoxMaterialHandle>,
)>::new(world);
let (box_mesh_handle, box_material_handle) =
system_state.get_mut(world).unwrap();
(box_mesh_handle.clone(), box_material_handle.clone())
};
world
.entity_mut(entity)
// Add our new `Mesh3d` and `MeshMaterial3d` to our tagged entity
.insert((
Mesh3d(box_mesh_handle),
MeshMaterial3d(box_material_handle),
transform,
));
});
command_queue
});
// Add our new task as a component
commands.entity(entity).insert(ComputeTransform(task));
}
}
}
}
/// This system queries for entities that have the `ComputeTransform` component.
/// It checks if the tasks associated with those entities are complete.
/// If the task is complete, it extracts the result, adds a new [`Mesh3d`] and [`MeshMaterial3d`]
/// to the entity using the result from the task, and removes the task component from the entity.
///
/// **Important Note:**
/// - Don't use `future::block_on(poll_once)` to check if tasks are completed, as it is expensive and
/// can block the main thread. Also, it leaves around a `Task<T>` which will panic if awaited again.
/// - Instead, use `check_ready` for efficient polling, which does not block the main thread.
fn handle_tasks(
mut commands: Commands,
mut transform_tasks: Query<(Entity, &mut ComputeTransform)>,
) {
for (entity, mut task) in &mut transform_tasks {
// Use `check_ready` to efficiently poll the task without blocking the main thread.
if let Some(mut commands_queue) = check_ready(&mut task.0) {
// Append the returned command queue to execute it later.
commands.append(&mut commands_queue);
// Task is complete, so remove the task component from the entity.
commands.entity(entity).remove::<ComputeTransform>();
}
}
}
/// This system is only used to setup light and camera for the environment
fn setup_env(mut commands: Commands) {
// Used to center camera on spawned cubes
let offset = if NUM_CUBES.is_multiple_of(2) {
(NUM_CUBES / 2) as f32 - 0.5
} else {
(NUM_CUBES / 2) as f32
};
// lights
commands.spawn((PointLight::default(), Transform::from_xyz(4.0, 12.0, 15.0)));
// camera
commands.spawn((
Camera3d::default(),
Transform::from_xyz(offset, offset, 15.0)
.looking_at(Vec3::new(offset, offset, 0.0), Vec3::Y),
));
}