Struct World

pub struct World { /* private fields */ }

Stores and exposes operations on entities, components, resources, and their associated metadata.

Each Entity has a set of components. Each component can have up to one instance of each component type. Entity components can be created, updated, removed, and queried using a given World.

For complex access patterns involving SystemParam, consider using SystemState.

To mutate different parts of the world simultaneously, use World::resource_scope or SystemState.

Resources

Worlds can also store Resources, which are unique instances of a given type that don’t belong to a specific Entity. There are also non send resources, which can only be accessed on the main thread. See Resource for usage.

Implementations

impl World

pub fn add_observer<E, B, M>( &mut self, system: impl IntoObserverSystem<E, B, M>, ) -> EntityWorldMut<'_>

where E: Event, B: Bundle,

Spawns a “global” Observer which will watch for the given event. Returns its Entity as a EntityWorldMut.

Calling observe on the returned EntityWorldMut will observe the observer itself, which you very likely do not want.

Example

#[derive(Component)]
struct A;

world.add_observer(|_: Trigger<OnAdd, A>| {
    // ...
});
world.add_observer(|_: Trigger<OnRemove, A>| {
    // ...
});

pub fn trigger(&mut self, event: impl Event)

Triggers the given Event, which will run any Observers watching for it.

While event types commonly implement Copy, those that don’t will be consumed and will no longer be accessible. If you need to use the event after triggering it, use World::trigger_ref instead.

pub fn trigger_ref(&mut self, event: &mut impl Event)

Triggers the given Event as a mutable reference, which will run any Observers watching for it.

Compared to World::trigger, this method is most useful when it’s necessary to check or use the event after it has been modified by observers.

pub fn trigger_targets( &mut self, event: impl Event, targets: impl TriggerTargets, )

Triggers the given Event for the given targets, which will run any Observers watching for it.

While event types commonly implement Copy, those that don’t will be consumed and will no longer be accessible. If you need to use the event after triggering it, use World::trigger_targets_ref instead.

pub fn trigger_targets_ref( &mut self, event: &mut impl Event, targets: impl TriggerTargets, )

Triggers the given Event as a mutable reference for the given targets, which will run any Observers watching for it.

Compared to World::trigger_targets, this method is most useful when it’s necessary to check or use the event after it has been modified by observers.

impl World

pub fn register_system<I, O, M>( &mut self, system: impl IntoSystem<I, O, M> + 'static, ) -> SystemId<I, O>

where I: SystemInput + 'static, O: 'static,

Registers a system and returns a SystemId so it can later be called by World::run_system.

It’s possible to register multiple copies of the same system by calling this function multiple times. If that’s not what you want, consider using World::register_system_cached instead.

This is different from adding systems to a Schedule, because the SystemId that is returned can be used anywhere in the World to run the associated system. This allows for running systems in a pushed-based fashion. Using a Schedule is still preferred for most cases due to its better performance and ability to run non-conflicting systems simultaneously.

Examples found in repository

examples/ecs/one_shot_systems.rs (line 51)

fn setup_with_world(world: &mut World) {
    // We can run it once manually
    world.run_system_once(system_b).unwrap();
    // Or with a Callback
    let system_id = world.register_system(system_b);
    world.spawn((Callback(system_id), B));
}

pub fn register_boxed_system<I, O>( &mut self, system: Box<dyn System<In = I, Out = O>>, ) -> SystemId<I, O>

where I: SystemInput + 'static, O: 'static,

Similar to Self::register_system, but allows passing in a BoxedSystem.

This is useful if the IntoSystem implementor has already been turned into a System trait object and put in a Box.

pub fn unregister_system<I, O>( &mut self, id: SystemId<I, O>, ) -> Result<RemovedSystem<I, O>, RegisteredSystemError<I, O>>

where I: SystemInput + 'static, O: 'static,

Removes a registered system and returns the system, if it exists. After removing a system, the SystemId becomes invalid and attempting to use it afterwards will result in errors. Re-adding the removed system will register it on a new SystemId.

If no system corresponds to the given SystemId, this method returns an error. Systems are also not allowed to remove themselves, this returns an error too.

pub fn run_system<O>( &mut self, id: SystemId<(), O>, ) -> Result<O, RegisteredSystemError<(), O>>

where O: 'static,

Run stored systems by their SystemId. Before running a system, it must first be registered. The method World::register_system stores a given system and returns a SystemId. This is different from RunSystemOnce::run_system_once, because it keeps local state between calls and change detection works correctly.

In order to run a chained system with an input, use World::run_system_with_input instead.

Limitations

• Stored systems cannot be recursive, they cannot call themselves through Commands::run_system.

Examples

Running a system

fn increment(mut counter: Local<u8>) {
   *counter += 1;
   println!("{}", *counter);
}

let mut world = World::default();
let counter_one = world.register_system(increment);
let counter_two = world.register_system(increment);
world.run_system(counter_one); // -> 1
world.run_system(counter_one); // -> 2
world.run_system(counter_two); // -> 1

Change detection

#[derive(Resource, Default)]
struct ChangeDetector;

let mut world = World::default();
world.init_resource::<ChangeDetector>();
let detector = world.register_system(|change_detector: ResMut<ChangeDetector>| {
    if change_detector.is_changed() {
        println!("Something happened!");
    } else {
        println!("Nothing happened.");
    }
});

// Resources are changed when they are first added
let _ = world.run_system(detector); // -> Something happened!
let _ = world.run_system(detector); // -> Nothing happened.
world.resource_mut::<ChangeDetector>().set_changed();
let _ = world.run_system(detector); // -> Something happened!

Getting system output

#[derive(Resource)]
struct PlayerScore(i32);

#[derive(Resource)]
struct OpponentScore(i32);

fn get_player_score(player_score: Res<PlayerScore>) -> i32 {
  player_score.0
}

fn get_opponent_score(opponent_score: Res<OpponentScore>) -> i32 {
  opponent_score.0
}

let mut world = World::default();
world.insert_resource(PlayerScore(3));
world.insert_resource(OpponentScore(2));

let scoring_systems = [
  ("player", world.register_system(get_player_score)),
  ("opponent", world.register_system(get_opponent_score)),
];

for (label, scoring_system) in scoring_systems {
  println!("{label} has score {}", world.run_system(scoring_system).expect("system succeeded"));
}

pub fn run_system_with_input<I, O>( &mut self, id: SystemId<I, O>, input: <I as SystemInput>::Inner<'_>, ) -> Result<O, RegisteredSystemError<I, O>>

where I: SystemInput + 'static, O: 'static,

Run a stored chained system by its SystemId, providing an input value. Before running a system, it must first be registered. The method World::register_system stores a given system and returns a SystemId.

Limitations

• Stored systems cannot be recursive, they cannot call themselves through Commands::run_system.

Examples

fn increment(In(increment_by): In<u8>, mut counter: Local<u8>) -> u8 {
  *counter += increment_by;
  *counter
}

let mut world = World::default();
let counter_one = world.register_system(increment);
let counter_two = world.register_system(increment);
assert_eq!(world.run_system_with_input(counter_one, 1).unwrap(), 1);
assert_eq!(world.run_system_with_input(counter_one, 20).unwrap(), 21);
assert_eq!(world.run_system_with_input(counter_two, 30).unwrap(), 30);

See World::run_system for more examples.

pub fn register_system_cached<I, O, M, S>( &mut self, system: S, ) -> SystemId<I, O>

where I: SystemInput + 'static, O: 'static, S: IntoSystem<I, O, M> + 'static,

Registers a system or returns its cached SystemId.

If you want to run the system immediately and you don’t need its SystemId, see World::run_system_cached.

The first time this function is called for a particular system, it will register it and store its SystemId in a CachedSystemId resource for later. If you would rather manage the SystemId yourself, or register multiple copies of the same system, use World::register_system instead.

Limitations

This function only accepts ZST (zero-sized) systems to guarantee that any two systems of the same type must be equal. This means that closures that capture the environment, and function pointers, are not accepted.

If you want to access values from the environment within a system, consider passing them in as inputs via World::run_system_cached_with. If that’s not an option, consider World::register_system instead.

pub fn unregister_system_cached<I, O, M, S>( &mut self, _system: S, ) -> Result<RemovedSystem<I, O>, RegisteredSystemError<I, O>>

where I: SystemInput + 'static, O: 'static, S: IntoSystem<I, O, M> + 'static,

Removes a cached system and its CachedSystemId resource.

See World::register_system_cached for more information.

pub fn run_system_cached<O, M, S>( &mut self, system: S, ) -> Result<O, RegisteredSystemError<(), O>>

where O: 'static, S: IntoSystem<(), O, M> + 'static,

Runs a cached system, registering it if necessary.

See World::register_system_cached for more information.

pub fn run_system_cached_with<I, O, M, S>( &mut self, system: S, input: <I as SystemInput>::Inner<'_>, ) -> Result<O, RegisteredSystemError<I, O>>

where I: SystemInput + 'static, O: 'static, S: IntoSystem<I, O, M> + 'static,

Runs a cached system with an input, registering it if necessary.

See World::register_system_cached for more information.

impl World

pub fn get_reflect( &self, entity: Entity, type_id: TypeId, ) -> Result<&(dyn Reflect + 'static), GetComponentReflectError>

Retrieves a reference to the given entity’s Component of the given type_id using reflection.

Requires implementing Reflect for the Component (e.g., using #[derive(Reflect)) and app.register_type::<TheComponent>() to have been called¹.

If you want to call this with a ComponentId, see World::components and Components::get_id to get the corresponding TypeId.

Also see the crate documentation for bevy_reflect for more information on Reflect and bevy’s reflection capabilities.

Errors

See GetComponentReflectError for the possible errors and their descriptions.

Example

use bevy_ecs::prelude::*;
use bevy_reflect::Reflect;
use std::any::TypeId;

// define a `Component` and derive `Reflect` for it
#[derive(Component, Reflect)]
struct MyComponent;

// create a `World` for this example
let mut world = World::new();

// Note: This is usually handled by `App::register_type()`, but this example cannot use `App`.
world.init_resource::<AppTypeRegistry>();
world.get_resource_mut::<AppTypeRegistry>().unwrap().write().register::<MyComponent>();

// spawn an entity with a `MyComponent`
let entity = world.spawn(MyComponent).id();

// retrieve a reflected reference to the entity's `MyComponent`
let comp_reflected: &dyn Reflect = world.get_reflect(entity, TypeId::of::<MyComponent>()).unwrap();

// make sure we got the expected type
assert!(comp_reflected.is::<MyComponent>());

Note

Requires the bevy_reflect feature (included in the default features).

---

1. More specifically: Requires TypeData for ReflectFromPtr to be registered for the given type_id, which is automatically handled when deriving Reflect and calling App::register_type. ↩

pub fn get_reflect_mut( &mut self, entity: Entity, type_id: TypeId, ) -> Result<Mut<'_, dyn Reflect>, GetComponentReflectError>

Retrieves a mutable reference to the given entity’s Component of the given type_id using reflection.

Requires implementing Reflect for the Component (e.g., using #[derive(Reflect)) and app.register_type::<TheComponent>() to have been called.

This is the mutable version of World::get_reflect, see its docs for more information and an example.

Just calling this method does not trigger change detection.

Errors

See GetComponentReflectError for the possible errors and their descriptions.

Example

See the documentation for World::get_reflect.

Note

Requires the feature bevy_reflect (included in the default features).

impl World

pub fn new() -> World

Creates a new empty World.

Panics

If usize::MAX Worlds have been created. This guarantee allows System Parameters to safely uniquely identify a World, since its WorldId is unique

Examples found in repository

examples/scene/scene.rs (line 99)

fn save_scene_system(world: &mut World) {
    // Scenes can be created from any ECS World.
    // You can either create a new one for the scene or use the current World.
    // For demonstration purposes, we'll create a new one.
    let mut scene_world = World::new();

    // The `TypeRegistry` resource contains information about all registered types (including components).
    // This is used to construct scenes, so we'll want to ensure that our previous type registrations
    // exist in this new scene world as well.
    // To do this, we can simply clone the `AppTypeRegistry` resource.
    let type_registry = world.resource::<AppTypeRegistry>().clone();
    scene_world.insert_resource(type_registry);

    let mut component_b = ComponentB::from_world(world);
    component_b.value = "hello".to_string();
    scene_world.spawn((
        component_b,
        ComponentA { x: 1.0, y: 2.0 },
        Transform::IDENTITY,
        Name::new("joe"),
    ));
    scene_world.spawn(ComponentA { x: 3.0, y: 4.0 });
    scene_world.insert_resource(ResourceA { score: 1 });

    // With our sample world ready to go, we can now create our scene using DynamicScene or DynamicSceneBuilder.
    // For simplicity, we will create our scene using DynamicScene:
    let scene = DynamicScene::from_world(&scene_world);

    // Scenes can be serialized like this:
    let type_registry = world.resource::<AppTypeRegistry>();
    let type_registry = type_registry.read();
    let serialized_scene = scene.serialize(&type_registry).unwrap();

    // Showing the scene in the console
    info!("{}", serialized_scene);

    // Writing the scene to a new file. Using a task to avoid calling the filesystem APIs in a system
    // as they are blocking
    // This can't work in Wasm as there is no filesystem access
    #[cfg(not(target_arch = "wasm32"))]
    IoTaskPool::get()
        .spawn(async move {
            // Write the scene RON data to file
            File::create(format!("assets/{NEW_SCENE_FILE_PATH}"))
                .and_then(|mut file| file.write(serialized_scene.as_bytes()))
                .expect("Error while writing scene to file");
        })
        .detach();
}

examples/ecs/dynamic.rs (line 48)

fn main() {
    let mut world = World::new();
    let mut lines = std::io::stdin().lines();
    let mut component_names = HashMap::<String, ComponentId>::new();
    let mut component_info = HashMap::<ComponentId, ComponentInfo>::new();

    println!("{PROMPT}");
    loop {
        print!("\n> ");
        let _ = std::io::stdout().flush();
        let Some(Ok(line)) = lines.next() else {
            return;
        };

        if line.is_empty() {
            return;
        };

        let Some((first, rest)) = line.trim().split_once(|c: char| c.is_whitespace()) else {
            match &line.chars().next() {
                Some('c') => println!("{COMPONENT_PROMPT}"),
                Some('s') => println!("{ENTITY_PROMPT}"),
                Some('q') => println!("{QUERY_PROMPT}"),
                _ => println!("{PROMPT}"),
            }
            continue;
        };

        match &first[0..1] {
            "c" => {
                rest.split(',').for_each(|component| {
                    let mut component = component.split_whitespace();
                    let Some(name) = component.next() else {
                        return;
                    };
                    let size = match component.next().map(str::parse) {
                        Some(Ok(size)) => size,
                        _ => 0,
                    };
                    // Register our new component to the world with a layout specified by it's size
                    // SAFETY: [u64] is Send + Sync
                    let id = world.register_component_with_descriptor(unsafe {
                        ComponentDescriptor::new_with_layout(
                            name.to_string(),
                            StorageType::Table,
                            Layout::array::<u64>(size).unwrap(),
                            None,
                        )
                    });
                    let Some(info) = world.components().get_info(id) else {
                        return;
                    };
                    component_names.insert(name.to_string(), id);
                    component_info.insert(id, info.clone());
                    println!("Component {} created with id: {:?}", name, id.index());
                });
            }
            "s" => {
                let mut to_insert_ids = Vec::new();
                let mut to_insert_data = Vec::new();
                rest.split(',').for_each(|component| {
                    let mut component = component.split_whitespace();
                    let Some(name) = component.next() else {
                        return;
                    };

                    // Get the id for the component with the given name
                    let Some(&id) = component_names.get(name) else {
                        println!("Component {name} does not exist");
                        return;
                    };

                    // Calculate the length for the array based on the layout created for this component id
                    let info = world.components().get_info(id).unwrap();
                    let len = info.layout().size() / size_of::<u64>();
                    let mut values: Vec<u64> = component
                        .take(len)
                        .filter_map(|value| value.parse::<u64>().ok())
                        .collect();
                    values.resize(len, 0);

                    // Collect the id and array to be inserted onto our entity
                    to_insert_ids.push(id);
                    to_insert_data.push(values);
                });

                let mut entity = world.spawn_empty();

                // Construct an `OwningPtr` for each component in `to_insert_data`
                let to_insert_ptr = to_owning_ptrs(&mut to_insert_data);

                // SAFETY:
                // - Component ids have been taken from the same world
                // - Each array is created to the layout specified in the world
                unsafe {
                    entity.insert_by_ids(&to_insert_ids, to_insert_ptr.into_iter());
                }

                println!("Entity spawned with id: {:?}", entity.id());
            }
            "q" => {
                let mut builder = QueryBuilder::<FilteredEntityMut>::new(&mut world);
                parse_query(rest, &mut builder, &component_names);
                let mut query = builder.build();

                query.iter_mut(&mut world).for_each(|filtered_entity| {
                    #[allow(deprecated)]
                    let terms = filtered_entity
                        .access()
                        .component_reads_and_writes()
                        .0
                        .map(|id| {
                            let ptr = filtered_entity.get_by_id(id).unwrap();
                            let info = component_info.get(&id).unwrap();
                            let len = info.layout().size() / size_of::<u64>();

                            // SAFETY:
                            // - All components are created with layout [u64]
                            // - len is calculated from the component descriptor
                            let data = unsafe {
                                std::slice::from_raw_parts_mut(
                                    ptr.assert_unique().as_ptr().cast::<u64>(),
                                    len,
                                )
                            };

                            // If we have write access, increment each value once
                            if filtered_entity.access().has_component_write(id) {
                                data.iter_mut().for_each(|data| {
                                    *data += 1;
                                });
                            }

                            format!("{}: {:?}", info.name(), data[0..len].to_vec())
                        })
                        .collect::<Vec<_>>()
                        .join(", ");

                    println!("{:?}: {}", filtered_entity.id(), terms);
                });
            }
            _ => continue,
        }
    }
}

pub fn id(&self) -> WorldId

Retrieves this World’s unique ID

pub fn as_unsafe_world_cell(&mut self) -> UnsafeWorldCell<'_>

Creates a new UnsafeWorldCell view with complete read+write access.

pub fn as_unsafe_world_cell_readonly(&self) -> UnsafeWorldCell<'_>

Creates a new UnsafeWorldCell view with only read access to everything.

pub fn entities(&self) -> &Entities

Retrieves this world’s Entities collection.

pub unsafe fn entities_mut(&mut self) -> &mut Entities

Retrieves this world’s Entities collection mutably.

Safety

Mutable reference must not be used to put the Entities data in an invalid state for this World

pub fn archetypes(&self) -> &Archetypes

Retrieves this world’s Archetypes collection.

pub fn components(&self) -> &Components

Retrieves this world’s Components collection.

Examples found in repository

examples/ecs/dynamic.rs (line 96)

fn main() {
    let mut world = World::new();
    let mut lines = std::io::stdin().lines();
    let mut component_names = HashMap::<String, ComponentId>::new();
    let mut component_info = HashMap::<ComponentId, ComponentInfo>::new();

    println!("{PROMPT}");
    loop {
        print!("\n> ");
        let _ = std::io::stdout().flush();
        let Some(Ok(line)) = lines.next() else {
            return;
        };

        if line.is_empty() {
            return;
        };

        let Some((first, rest)) = line.trim().split_once(|c: char| c.is_whitespace()) else {
            match &line.chars().next() {
                Some('c') => println!("{COMPONENT_PROMPT}"),
                Some('s') => println!("{ENTITY_PROMPT}"),
                Some('q') => println!("{QUERY_PROMPT}"),
                _ => println!("{PROMPT}"),
            }
            continue;
        };

        match &first[0..1] {
            "c" => {
                rest.split(',').for_each(|component| {
                    let mut component = component.split_whitespace();
                    let Some(name) = component.next() else {
                        return;
                    };
                    let size = match component.next().map(str::parse) {
                        Some(Ok(size)) => size,
                        _ => 0,
                    };
                    // Register our new component to the world with a layout specified by it's size
                    // SAFETY: [u64] is Send + Sync
                    let id = world.register_component_with_descriptor(unsafe {
                        ComponentDescriptor::new_with_layout(
                            name.to_string(),
                            StorageType::Table,
                            Layout::array::<u64>(size).unwrap(),
                            None,
                        )
                    });
                    let Some(info) = world.components().get_info(id) else {
                        return;
                    };
                    component_names.insert(name.to_string(), id);
                    component_info.insert(id, info.clone());
                    println!("Component {} created with id: {:?}", name, id.index());
                });
            }
            "s" => {
                let mut to_insert_ids = Vec::new();
                let mut to_insert_data = Vec::new();
                rest.split(',').for_each(|component| {
                    let mut component = component.split_whitespace();
                    let Some(name) = component.next() else {
                        return;
                    };

                    // Get the id for the component with the given name
                    let Some(&id) = component_names.get(name) else {
                        println!("Component {name} does not exist");
                        return;
                    };

                    // Calculate the length for the array based on the layout created for this component id
                    let info = world.components().get_info(id).unwrap();
                    let len = info.layout().size() / size_of::<u64>();
                    let mut values: Vec<u64> = component
                        .take(len)
                        .filter_map(|value| value.parse::<u64>().ok())
                        .collect();
                    values.resize(len, 0);

                    // Collect the id and array to be inserted onto our entity
                    to_insert_ids.push(id);
                    to_insert_data.push(values);
                });

                let mut entity = world.spawn_empty();

                // Construct an `OwningPtr` for each component in `to_insert_data`
                let to_insert_ptr = to_owning_ptrs(&mut to_insert_data);

                // SAFETY:
                // - Component ids have been taken from the same world
                // - Each array is created to the layout specified in the world
                unsafe {
                    entity.insert_by_ids(&to_insert_ids, to_insert_ptr.into_iter());
                }

                println!("Entity spawned with id: {:?}", entity.id());
            }
            "q" => {
                let mut builder = QueryBuilder::<FilteredEntityMut>::new(&mut world);
                parse_query(rest, &mut builder, &component_names);
                let mut query = builder.build();

                query.iter_mut(&mut world).for_each(|filtered_entity| {
                    #[allow(deprecated)]
                    let terms = filtered_entity
                        .access()
                        .component_reads_and_writes()
                        .0
                        .map(|id| {
                            let ptr = filtered_entity.get_by_id(id).unwrap();
                            let info = component_info.get(&id).unwrap();
                            let len = info.layout().size() / size_of::<u64>();

                            // SAFETY:
                            // - All components are created with layout [u64]
                            // - len is calculated from the component descriptor
                            let data = unsafe {
                                std::slice::from_raw_parts_mut(
                                    ptr.assert_unique().as_ptr().cast::<u64>(),
                                    len,
                                )
                            };

                            // If we have write access, increment each value once
                            if filtered_entity.access().has_component_write(id) {
                                data.iter_mut().for_each(|data| {
                                    *data += 1;
                                });
                            }

                            format!("{}: {:?}", info.name(), data[0..len].to_vec())
                        })
                        .collect::<Vec<_>>()
                        .join(", ");

                    println!("{:?}: {}", filtered_entity.id(), terms);
                });
            }
            _ => continue,
        }
    }
}

pub fn storages(&self) -> &Storages

Retrieves this world’s Storages collection.

pub fn bundles(&self) -> &Bundles

Retrieves this world’s Bundles collection.

pub fn removed_components(&self) -> &RemovedComponentEvents

Retrieves this world’s RemovedComponentEvents collection

pub fn commands(&mut self) -> Commands<'_, '_>

Creates a new Commands instance that writes to the world’s command queue Use World::flush to apply all queued commands

pub fn register_component<T>(&mut self) -> ComponentId

where T: Component,

Registers a new Component type and returns the ComponentId created for it.

pub fn register_component_hooks<T>(&mut self) -> &mut ComponentHooks

where T: Component,

Returns a mutable reference to the ComponentHooks for a Component type.

Will panic if T exists in any archetypes.

Examples found in repository

examples/ecs/component_hooks.rs (line 63)

fn setup(world: &mut World) {
    // In order to register component hooks the component must:
    // - not be currently in use by any entities in the world
    // - not already have a hook of that kind registered
    // This is to prevent overriding hooks defined in plugins and other crates as well as keeping things fast
    world
        .register_component_hooks::<MyComponent>()
        // There are 4 component lifecycle hooks: `on_add`, `on_insert`, `on_replace` and `on_remove`
        // A hook has 3 arguments:
        // - a `DeferredWorld`, this allows access to resource and component data as well as `Commands`
        // - the entity that triggered the hook
        // - the component id of the triggering component, this is mostly used for dynamic components
        //
        // `on_add` will trigger when a component is inserted onto an entity without it
        .on_add(|mut world, entity, component_id| {
            // You can access component data from within the hook
            let value = world.get::<MyComponent>(entity).unwrap().0;
            println!("Component: {component_id:?} added to: {entity:?} with value {value:?}");
            // Or access resources
            world
                .resource_mut::<MyComponentIndex>()
                .insert(value, entity);
            // Or send events
            world.send_event(MyEvent);
        })
        // `on_insert` will trigger when a component is inserted onto an entity,
        // regardless of whether or not it already had it and after `on_add` if it ran
        .on_insert(|world, _, _| {
            println!("Current Index: {:?}", world.resource::<MyComponentIndex>());
        })
        // `on_replace` will trigger when a component is inserted onto an entity that already had it,
        // and runs before the value is replaced.
        // Also triggers when a component is removed from an entity, and runs before `on_remove`
        .on_replace(|mut world, entity, _| {
            let value = world.get::<MyComponent>(entity).unwrap().0;
            world.resource_mut::<MyComponentIndex>().remove(&value);
        })
        // `on_remove` will trigger when a component is removed from an entity,
        // since it runs before the component is removed you can still access the component data
        .on_remove(|mut world, entity, component_id| {
            let value = world.get::<MyComponent>(entity).unwrap().0;
            println!("Component: {component_id:?} removed from: {entity:?} with value {value:?}");
            // You can also issue commands through `.commands()`
            world.commands().entity(entity).despawn();
        });
}

pub fn register_component_hooks_by_id( &mut self, id: ComponentId, ) -> Option<&mut ComponentHooks>

Returns a mutable reference to the ComponentHooks for a Component with the given id if it exists.

Will panic if id exists in any archetypes.

pub fn register_required_components<T, R>(&mut self)

where T: Component, R: Component + Default,

Registers the given component R as a required component for T.

When T is added to an entity, R and its own required components will also be added if R was not already provided. The Default constructor will be used for the creation of R. If a custom constructor is desired, use World::register_required_components_with instead.

For the non-panicking version, see World::try_register_required_components.

Note that requirements must currently be registered before T is inserted into the world for the first time. This limitation may be fixed in the future.

Panics

Panics if R is already a directly required component for T, or if T has ever been added on an entity before the registration.

Indirect requirements through other components are allowed. In those cases, any existing requirements will only be overwritten if the new requirement is more specific.

Example

#[derive(Component)]
struct A;

#[derive(Component, Default, PartialEq, Eq, Debug)]
struct B(usize);

#[derive(Component, Default, PartialEq, Eq, Debug)]
struct C(u32);

// Register B as required by A and C as required by B.
world.register_required_components::<A, B>();
world.register_required_components::<B, C>();

// This will implicitly also insert B and C with their Default constructors.
let id = world.spawn(A).id();
assert_eq!(&B(0), world.entity(id).get::<B>().unwrap());
assert_eq!(&C(0), world.entity(id).get::<C>().unwrap());

pub fn register_required_components_with<T, R>( &mut self, constructor: fn() -> R, )

where T: Component, R: Component,

Registers the given component R as a required component for T.

When T is added to an entity, R and its own required components will also be added if R was not already provided. The given constructor will be used for the creation of R. If a Default constructor is desired, use World::register_required_components instead.

For the non-panicking version, see World::try_register_required_components_with.

Note that requirements must currently be registered before T is inserted into the world for the first time. This limitation may be fixed in the future.

Panics

Panics if R is already a directly required component for T, or if T has ever been added on an entity before the registration.

Indirect requirements through other components are allowed. In those cases, any existing requirements will only be overwritten if the new requirement is more specific.

Example

#[derive(Component)]
struct A;

#[derive(Component, Default, PartialEq, Eq, Debug)]
struct B(usize);

#[derive(Component, PartialEq, Eq, Debug)]
struct C(u32);

// Register B and C as required by A and C as required by B.
// A requiring C directly will overwrite the indirect requirement through B.
world.register_required_components::<A, B>();
world.register_required_components_with::<B, C>(|| C(1));
world.register_required_components_with::<A, C>(|| C(2));

// This will implicitly also insert B with its Default constructor and C
// with the custom constructor defined by A.
let id = world.spawn(A).id();
assert_eq!(&B(0), world.entity(id).get::<B>().unwrap());
assert_eq!(&C(2), world.entity(id).get::<C>().unwrap());

pub fn try_register_required_components<T, R>( &mut self, ) -> Result<(), RequiredComponentsError>

where T: Component, R: Component + Default,

Tries to register the given component R as a required component for T.

When T is added to an entity, R and its own required components will also be added if R was not already provided. The Default constructor will be used for the creation of R. If a custom constructor is desired, use World::register_required_components_with instead.

For the panicking version, see World::register_required_components.

Note that requirements must currently be registered before T is inserted into the world for the first time. This limitation may be fixed in the future.

Errors

Returns a RequiredComponentsError if R is already a directly required component for T, or if T has ever been added on an entity before the registration.

Indirect requirements through other components are allowed. In those cases, any existing requirements will only be overwritten if the new requirement is more specific.

Example

#[derive(Component)]
struct A;

#[derive(Component, Default, PartialEq, Eq, Debug)]
struct B(usize);

#[derive(Component, Default, PartialEq, Eq, Debug)]
struct C(u32);

// Register B as required by A and C as required by B.
world.register_required_components::<A, B>();
world.register_required_components::<B, C>();

// Duplicate registration! This will fail.
assert!(world.try_register_required_components::<A, B>().is_err());

// This will implicitly also insert B and C with their Default constructors.
let id = world.spawn(A).id();
assert_eq!(&B(0), world.entity(id).get::<B>().unwrap());
assert_eq!(&C(0), world.entity(id).get::<C>().unwrap());

pub fn try_register_required_components_with<T, R>( &mut self, constructor: fn() -> R, ) -> Result<(), RequiredComponentsError>

where T: Component, R: Component,

Tries to register the given component R as a required component for T.

When T is added to an entity, R and its own required components will also be added if R was not already provided. The given constructor will be used for the creation of R. If a Default constructor is desired, use World::register_required_components instead.

For the panicking version, see World::register_required_components_with.

Note that requirements must currently be registered before T is inserted into the world for the first time. This limitation may be fixed in the future.

Errors

Returns a RequiredComponentsError if R is already a directly required component for T, or if T has ever been added on an entity before the registration.

Indirect requirements through other components are allowed. In those cases, any existing requirements will only be overwritten if the new requirement is more specific.

Example

#[derive(Component)]
struct A;

#[derive(Component, Default, PartialEq, Eq, Debug)]
struct B(usize);

#[derive(Component, PartialEq, Eq, Debug)]
struct C(u32);

// Register B and C as required by A and C as required by B.
// A requiring C directly will overwrite the indirect requirement through B.
world.register_required_components::<A, B>();
world.register_required_components_with::<B, C>(|| C(1));
world.register_required_components_with::<A, C>(|| C(2));

// Duplicate registration! Even if the constructors were different, this would fail.
assert!(world.try_register_required_components_with::<B, C>(|| C(1)).is_err());

// This will implicitly also insert B with its Default constructor and C
// with the custom constructor defined by A.
let id = world.spawn(A).id();
assert_eq!(&B(0), world.entity(id).get::<B>().unwrap());
assert_eq!(&C(2), world.entity(id).get::<C>().unwrap());

pub fn get_required_components<C>(&self) -> Option<&RequiredComponents>

where C: Component,

Retrieves the required components for the given component type, if it exists.

pub fn get_required_components_by_id( &self, id: ComponentId, ) -> Option<&RequiredComponents>

Retrieves the required components for the component of the given ComponentId, if it exists.

pub fn register_component_with_descriptor( &mut self, descriptor: ComponentDescriptor, ) -> ComponentId

Registers a new Component type and returns the ComponentId created for it.

This method differs from World::register_component in that it uses a ComponentDescriptor to register the new component type instead of statically available type information. This enables the dynamic registration of new component definitions at runtime for advanced use cases.

While the option to register a component from a descriptor is useful in type-erased contexts, the standard World::register_component function should always be used instead when type information is available at compile time.

Examples found in repository

examples/ecs/dynamic.rs (lines 88-95)

fn main() {
    let mut world = World::new();
    let mut lines = std::io::stdin().lines();
    let mut component_names = HashMap::<String, ComponentId>::new();
    let mut component_info = HashMap::<ComponentId, ComponentInfo>::new();

    println!("{PROMPT}");
    loop {
        print!("\n> ");
        let _ = std::io::stdout().flush();
        let Some(Ok(line)) = lines.next() else {
            return;
        };

        if line.is_empty() {
            return;
        };

        let Some((first, rest)) = line.trim().split_once(|c: char| c.is_whitespace()) else {
            match &line.chars().next() {
                Some('c') => println!("{COMPONENT_PROMPT}"),
                Some('s') => println!("{ENTITY_PROMPT}"),
                Some('q') => println!("{QUERY_PROMPT}"),
                _ => println!("{PROMPT}"),
            }
            continue;
        };

        match &first[0..1] {
            "c" => {
                rest.split(',').for_each(|component| {
                    let mut component = component.split_whitespace();
                    let Some(name) = component.next() else {
                        return;
                    };
                    let size = match component.next().map(str::parse) {
                        Some(Ok(size)) => size,
                        _ => 0,
                    };
                    // Register our new component to the world with a layout specified by it's size
                    // SAFETY: [u64] is Send + Sync
                    let id = world.register_component_with_descriptor(unsafe {
                        ComponentDescriptor::new_with_layout(
                            name.to_string(),
                            StorageType::Table,
                            Layout::array::<u64>(size).unwrap(),
                            None,
                        )
                    });
                    let Some(info) = world.components().get_info(id) else {
                        return;
                    };
                    component_names.insert(name.to_string(), id);
                    component_info.insert(id, info.clone());
                    println!("Component {} created with id: {:?}", name, id.index());
                });
            }
            "s" => {
                let mut to_insert_ids = Vec::new();
                let mut to_insert_data = Vec::new();
                rest.split(',').for_each(|component| {
                    let mut component = component.split_whitespace();
                    let Some(name) = component.next() else {
                        return;
                    };

                    // Get the id for the component with the given name
                    let Some(&id) = component_names.get(name) else {
                        println!("Component {name} does not exist");
                        return;
                    };

                    // Calculate the length for the array based on the layout created for this component id
                    let info = world.components().get_info(id).unwrap();
                    let len = info.layout().size() / size_of::<u64>();
                    let mut values: Vec<u64> = component
                        .take(len)
                        .filter_map(|value| value.parse::<u64>().ok())
                        .collect();
                    values.resize(len, 0);

                    // Collect the id and array to be inserted onto our entity
                    to_insert_ids.push(id);
                    to_insert_data.push(values);
                });

                let mut entity = world.spawn_empty();

                // Construct an `OwningPtr` for each component in `to_insert_data`
                let to_insert_ptr = to_owning_ptrs(&mut to_insert_data);

                // SAFETY:
                // - Component ids have been taken from the same world
                // - Each array is created to the layout specified in the world
                unsafe {
                    entity.insert_by_ids(&to_insert_ids, to_insert_ptr.into_iter());
                }

                println!("Entity spawned with id: {:?}", entity.id());
            }
            "q" => {
                let mut builder = QueryBuilder::<FilteredEntityMut>::new(&mut world);
                parse_query(rest, &mut builder, &component_names);
                let mut query = builder.build();

                query.iter_mut(&mut world).for_each(|filtered_entity| {
                    #[allow(deprecated)]
                    let terms = filtered_entity
                        .access()
                        .component_reads_and_writes()
                        .0
                        .map(|id| {
                            let ptr = filtered_entity.get_by_id(id).unwrap();
                            let info = component_info.get(&id).unwrap();
                            let len = info.layout().size() / size_of::<u64>();

                            // SAFETY:
                            // - All components are created with layout [u64]
                            // - len is calculated from the component descriptor
                            let data = unsafe {
                                std::slice::from_raw_parts_mut(
                                    ptr.assert_unique().as_ptr().cast::<u64>(),
                                    len,
                                )
                            };

                            // If we have write access, increment each value once
                            if filtered_entity.access().has_component_write(id) {
                                data.iter_mut().for_each(|data| {
                                    *data += 1;
                                });
                            }

                            format!("{}: {:?}", info.name(), data[0..len].to_vec())
                        })
                        .collect::<Vec<_>>()
                        .join(", ");

                    println!("{:?}: {}", filtered_entity.id(), terms);
                });
            }
            _ => continue,
        }
    }
}

pub fn component_id<T>(&self) -> Option<ComponentId>

where T: Component,

Returns the ComponentId of the given Component type T.

The returned ComponentId is specific to the World instance it was retrieved from and should not be used with another World instance.

Returns None if the Component type has not yet been initialized within the World using World::register_component.

use bevy_ecs::prelude::*;

let mut world = World::new();

#[derive(Component)]
struct ComponentA;

let component_a_id = world.register_component::<ComponentA>();

assert_eq!(component_a_id, world.component_id::<ComponentA>().unwrap())

See also

• Components::component_id()
• Components::get_id()

pub fn register_resource<R>(&mut self) -> ComponentId

where R: Resource,

Registers a new Resource type and returns the ComponentId created for it.

The Resource doesn’t have a value in the World, it’s only registered. If you want to insert the Resource in the World, use World::init_resource or World::insert_resource instead.

pub fn resource_id<T>(&self) -> Option<ComponentId>

where T: Resource,

Returns the ComponentId of the given Resource type T.

The returned ComponentId is specific to the World instance it was retrieved from and should not be used with another World instance.

Returns None if the Resource type has not yet been initialized within the World using World::register_resource, World::init_resource or World::insert_resource.

pub fn entity<F>(&self, entities: F) -> <F as WorldEntityFetch>::Ref<'_>

where F: WorldEntityFetch,

Returns EntityRefs that expose read-only operations for the given entities. This will panic if any of the given entities do not exist. Use World::get_entity if you want to check for entity existence instead of implicitly panicking.

This function supports fetching a single entity or multiple entities:

• Pass an Entity to receive a single EntityRef.
• Pass a slice of Entitys to receive a Vec<EntityRef>.
• Pass an array of Entitys to receive an equally-sized array of EntityRefs.
• Pass a reference to a EntityHashSet to receive an EntityHashMap<EntityRef>.

Panics

If any of the given entities do not exist in the world.

Examples

Single Entity

#[derive(Component)]
struct Position {
  x: f32,
  y: f32,
}

let mut world = World::new();
let entity = world.spawn(Position { x: 0.0, y: 0.0 }).id();

let position = world.entity(entity).get::<Position>().unwrap();
assert_eq!(position.x, 0.0);

Array of Entitys

#[derive(Component)]
struct Position {
  x: f32,
  y: f32,
}

let mut world = World::new();
let e1 = world.spawn(Position { x: 0.0, y: 0.0 }).id();
let e2 = world.spawn(Position { x: 1.0, y: 1.0 }).id();

let [e1_ref, e2_ref] = world.entity([e1, e2]);
let e1_position = e1_ref.get::<Position>().unwrap();
assert_eq!(e1_position.x, 0.0);
let e2_position = e2_ref.get::<Position>().unwrap();
assert_eq!(e2_position.x, 1.0);

Slice of Entitys

#[derive(Component)]
struct Position {
  x: f32,
  y: f32,
}

let mut world = World::new();
let e1 = world.spawn(Position { x: 0.0, y: 1.0 }).id();
let e2 = world.spawn(Position { x: 0.0, y: 1.0 }).id();
let e3 = world.spawn(Position { x: 0.0, y: 1.0 }).id();

let ids = vec![e1, e2, e3];
for eref in world.entity(&ids[..]) {
    assert_eq!(eref.get::<Position>().unwrap().y, 1.0);
}

EntityHashSet

#[derive(Component)]
struct Position {
  x: f32,
  y: f32,
}

let mut world = World::new();
let e1 = world.spawn(Position { x: 0.0, y: 1.0 }).id();
let e2 = world.spawn(Position { x: 0.0, y: 1.0 }).id();
let e3 = world.spawn(Position { x: 0.0, y: 1.0 }).id();

let ids = EntityHashSet::from_iter([e1, e2, e3]);
for (_id, eref) in world.entity(&ids) {
    assert_eq!(eref.get::<Position>().unwrap().y, 1.0);
}

pub fn entity_mut<F>(&mut self, entities: F) -> <F as WorldEntityFetch>::Mut<'_>

where F: WorldEntityFetch,

Returns EntityMuts that expose read and write operations for the given entities. This will panic if any of the given entities do not exist. Use World::get_entity_mut if you want to check for entity existence instead of implicitly panicking.

This function supports fetching a single entity or multiple entities:

• Pass an Entity to receive a single EntityWorldMut.
  • This reference type allows for structural changes to the entity, such as adding or removing components, or despawning the entity.
• Pass a slice of Entitys to receive a Vec<EntityMut>.
• Pass an array of Entitys to receive an equally-sized array of EntityMuts.
• Pass a reference to a EntityHashSet to receive an EntityHashMap<EntityMut>.

In order to perform structural changes on the returned entity reference, such as adding or removing components, or despawning the entity, only a single Entity can be passed to this function. Allowing multiple entities at the same time with structural access would lead to undefined behavior, so EntityMut is returned when requesting multiple entities.

Panics

If any of the given entities do not exist in the world.

Examples

Single Entity

#[derive(Component)]
struct Position {
  x: f32,
  y: f32,
}

let mut world = World::new();
let entity = world.spawn(Position { x: 0.0, y: 0.0 }).id();

let mut entity_mut = world.entity_mut(entity);
let mut position = entity_mut.get_mut::<Position>().unwrap();
position.y = 1.0;
assert_eq!(position.x, 0.0);
entity_mut.despawn();

Array of Entitys

#[derive(Component)]
struct Position {
  x: f32,
  y: f32,
}

let mut world = World::new();
let e1 = world.spawn(Position { x: 0.0, y: 0.0 }).id();
let e2 = world.spawn(Position { x: 1.0, y: 1.0 }).id();

let [mut e1_ref, mut e2_ref] = world.entity_mut([e1, e2]);
let mut e1_position = e1_ref.get_mut::<Position>().unwrap();
e1_position.x = 1.0;
assert_eq!(e1_position.x, 1.0);
let mut e2_position = e2_ref.get_mut::<Position>().unwrap();
e2_position.x = 2.0;
assert_eq!(e2_position.x, 2.0);

Slice of Entitys

#[derive(Component)]
struct Position {
  x: f32,
  y: f32,
}

let mut world = World::new();
let e1 = world.spawn(Position { x: 0.0, y: 1.0 }).id();
let e2 = world.spawn(Position { x: 0.0, y: 1.0 }).id();
let e3 = world.spawn(Position { x: 0.0, y: 1.0 }).id();

let ids = vec![e1, e2, e3];
for mut eref in world.entity_mut(&ids[..]) {
    let mut pos = eref.get_mut::<Position>().unwrap();
    pos.y = 2.0;
    assert_eq!(pos.y, 2.0);
}

EntityHashSet

#[derive(Component)]
struct Position {
  x: f32,
  y: f32,
}

let mut world = World::new();
let e1 = world.spawn(Position { x: 0.0, y: 1.0 }).id();
let e2 = world.spawn(Position { x: 0.0, y: 1.0 }).id();
let e3 = world.spawn(Position { x: 0.0, y: 1.0 }).id();

let ids = EntityHashSet::from_iter([e1, e2, e3]);
for (_id, mut eref) in world.entity_mut(&ids) {
    let mut pos = eref.get_mut::<Position>().unwrap();
    pos.y = 2.0;
    assert_eq!(pos.y, 2.0);
}

Examples found in repository

examples/async_tasks/async_compute.rs (line 86)

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::thread_rng().gen_range(0.05..5.0));

                    // Pretend this is a time-intensive function. :)
                    async_std::task::sleep(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);

                            (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,
                            ))
                            // Task is complete, so remove task component from entity
                            .remove::<ComputeTransform>();
                    });

                    command_queue
                });

                // Spawn new entity and add our new task as a component
                commands.entity(entity).insert(ComputeTransform(task));
            }
        }
    }
}

pub fn many_entities<const N: usize>( &mut self, entities: [Entity; N], ) -> [EntityRef<'_>; N]

👎Deprecated since 0.15.0: Use World::entity::<[Entity; N]> instead

Gets an EntityRef for multiple entities at once.

Panics

If any entity does not exist in the world.

Examples

// Getting multiple entities.
let [entity1, entity2] = world.many_entities([id1, id2]);

// Trying to get a despawned entity will fail.
world.despawn(id2);
world.many_entities([id1, id2]);

pub fn many_entities_mut<const N: usize>( &mut self, entities: [Entity; N], ) -> [EntityMut<'_>; N]

👎Deprecated since 0.15.0: Use World::entity_mut::<[Entity; N]> instead

Gets mutable access to multiple entities at once.

Panics

If any entities do not exist in the world, or if the same entity is specified multiple times.

Examples

Disjoint mutable access.

// Disjoint mutable access.
let [entity1, entity2] = world.many_entities_mut([id1, id2]);

Trying to access the same entity multiple times will fail.

world.many_entities_mut([id, id]);

pub fn inspect_entity( &self, entity: Entity, ) -> impl Iterator<Item = &ComponentInfo>

Returns the components of an Entity through ComponentInfo.

pub fn get_or_spawn(&mut self, entity: Entity) -> Option<EntityWorldMut<'_>>

👎Deprecated since 0.15.0: use World::spawn instead

Returns an EntityWorldMut for the given entity (if it exists) or spawns one if it doesn’t exist. This will return None if the entity exists with a different generation.

Note

Spawning a specific entity value is rarely the right choice. Most apps should favor World::spawn. This method should generally only be used for sharing entities across apps, and only when they have a scheme worked out to share an ID space (which doesn’t happen by default).

pub fn get_entity<F>( &self, entities: F, ) -> Result<<F as WorldEntityFetch>::Ref<'_>, Entity>

where F: WorldEntityFetch,

Returns EntityRefs that expose read-only operations for the given entities, returning Err if any of the given entities do not exist. Instead of immediately unwrapping the value returned from this function, prefer World::entity.

This function supports fetching a single entity or multiple entities:

• Pass an Entity to receive a single EntityRef.
• Pass a slice of Entitys to receive a Vec<EntityRef>.
• Pass an array of Entitys to receive an equally-sized array of EntityRefs.
• Pass a reference to a EntityHashSet to receive an EntityHashMap<EntityRef>.

Errors

If any of the given entities do not exist in the world, the first Entity found to be missing will be returned in the Err.

Examples

For examples, see World::entity.

pub fn get_many_entities<const N: usize>( &self, entities: [Entity; N], ) -> Result<[EntityRef<'_>; N], Entity>

👎Deprecated since 0.15.0: Use World::get_entity::<[Entity; N]> instead

Gets an EntityRef for multiple entities at once.

Errors

If any entity does not exist in the world.

Examples

// Getting multiple entities.
let [entity1, entity2] = world.get_many_entities([id1, id2]).unwrap();

// Trying to get a despawned entity will fail.
world.despawn(id2);
assert!(world.get_many_entities([id1, id2]).is_err());

pub fn get_many_entities_dynamic<'w>( &'w self, entities: &[Entity], ) -> Result<Vec<EntityRef<'w>>, Entity>

👎Deprecated since 0.15.0: Use World::get_entity::<&[Entity]> instead

Gets an EntityRef for multiple entities at once, whose number is determined at runtime.

Errors

If any entity does not exist in the world.

Examples

// Getting multiple entities.
let entities = world.get_many_entities_dynamic(&[id1, id2]).unwrap();
let entity1 = entities.get(0).unwrap();
let entity2 = entities.get(1).unwrap();

// Trying to get a despawned entity will fail.
world.despawn(id2);
assert!(world.get_many_entities_dynamic(&[id1, id2]).is_err());

pub fn get_entity_mut<F>( &mut self, entities: F, ) -> Result<<F as WorldEntityFetch>::Mut<'_>, EntityFetchError>

where F: WorldEntityFetch,

Returns EntityMuts that expose read and write operations for the given entities, returning Err if any of the given entities do not exist. Instead of immediately unwrapping the value returned from this function, prefer World::entity_mut.

This function supports fetching a single entity or multiple entities:

• Pass an Entity to receive a single EntityWorldMut.
  • This reference type allows for structural changes to the entity, such as adding or removing components, or despawning the entity.
• Pass a slice of Entitys to receive a Vec<EntityMut>.
• Pass an array of Entitys to receive an equally-sized array of EntityMuts.
• Pass a reference to a EntityHashSet to receive an EntityHashMap<EntityMut>.

In order to perform structural changes on the returned entity reference, such as adding or removing components, or despawning the entity, only a single Entity can be passed to this function. Allowing multiple entities at the same time with structural access would lead to undefined behavior, so EntityMut is returned when requesting multiple entities.

Errors

• Returns EntityFetchError::NoSuchEntity if any of the given entities do not exist in the world.
  • Only the first entity found to be missing will be returned.
• Returns EntityFetchError::AliasedMutability if the same entity is requested multiple times.

Examples

For examples, see World::entity_mut.

pub fn iter_entities(&self) -> impl Iterator<Item = EntityRef<'_>>

Returns an Entity iterator of current entities.

This is useful in contexts where you only have read-only access to the World.

pub fn iter_entities_mut(&mut self) -> impl Iterator<Item = EntityMut<'_>>

Returns a mutable iterator over all entities in the World.

pub fn get_many_entities_mut<const N: usize>( &mut self, entities: [Entity; N], ) -> Result<[EntityMut<'_>; N], QueryEntityError<'_>>

👎Deprecated since 0.15.0: Use World::get_entity_mut::<[Entity; N]> instead

Gets mutable access to multiple entities.

Errors

If any entities do not exist in the world, or if the same entity is specified multiple times.

Examples

// Disjoint mutable access.
let [entity1, entity2] = world.get_many_entities_mut([id1, id2]).unwrap();

// Trying to access the same entity multiple times will fail.
assert!(world.get_many_entities_mut([id1, id1]).is_err());

pub fn get_many_entities_dynamic_mut<'w>( &'w mut self, entities: &[Entity], ) -> Result<Vec<EntityMut<'w>>, QueryEntityError<'w>>

👎Deprecated since 0.15.0: Use World::get_entity_mut::<&[Entity]> instead

Gets mutable access to multiple entities, whose number is determined at runtime.

Errors

If any entities do not exist in the world, or if the same entity is specified multiple times.

Examples

// Disjoint mutable access.
let mut entities = world.get_many_entities_dynamic_mut(&[id1, id2]).unwrap();
let entity1 = entities.get_mut(0).unwrap();

// Trying to access the same entity multiple times will fail.
assert!(world.get_many_entities_dynamic_mut(&[id1, id1]).is_err());

pub fn get_many_entities_from_set_mut<'w>( &'w mut self, entities: &HashSet<Entity, EntityHash>, ) -> Result<Vec<EntityMut<'w>>, QueryEntityError<'w>>

👎Deprecated since 0.15.0: Use World::get_entity_mut::<&EntityHashSet> instead.

Gets mutable access to multiple entities, contained in a EntityHashSet. The uniqueness of items in a EntityHashSet allows us to avoid checking for duplicates.

Errors

If any entities do not exist in the world.

Examples

let s = EntityHash::default();
let mut set = EntityHashSet::with_hasher(s);
set.insert(id1);
set.insert(id2);

// Disjoint mutable access.
let mut entities = world.get_many_entities_from_set_mut(&set).unwrap();
let entity1 = entities.get_mut(0).unwrap();

pub fn spawn_empty(&mut self) -> EntityWorldMut<'_>

Spawns a new Entity and returns a corresponding EntityWorldMut, which can be used to add components to the entity or retrieve its id.

use bevy_ecs::{component::Component, world::World};

#[derive(Component)]
struct Position {
  x: f32,
  y: f32,
}
#[derive(Component)]
struct Label(&'static str);
#[derive(Component)]
struct Num(u32);

let mut world = World::new();
let entity = world.spawn_empty()
    .insert(Position { x: 0.0, y: 0.0 }) // add a single component
    .insert((Num(1), Label("hello"))) // add a bundle of components
    .id();

let position = world.entity(entity).get::<Position>().unwrap();
assert_eq!(position.x, 0.0);

Examples found in repository

examples/ecs/dynamic.rs (line 133)

fn main() {
    let mut world = World::new();
    let mut lines = std::io::stdin().lines();
    let mut component_names = HashMap::<String, ComponentId>::new();
    let mut component_info = HashMap::<ComponentId, ComponentInfo>::new();

    println!("{PROMPT}");
    loop {
        print!("\n> ");
        let _ = std::io::stdout().flush();
        let Some(Ok(line)) = lines.next() else {
            return;
        };

        if line.is_empty() {
            return;
        };

        let Some((first, rest)) = line.trim().split_once(|c: char| c.is_whitespace()) else {
            match &line.chars().next() {
                Some('c') => println!("{COMPONENT_PROMPT}"),
                Some('s') => println!("{ENTITY_PROMPT}"),
                Some('q') => println!("{QUERY_PROMPT}"),
                _ => println!("{PROMPT}"),
            }
            continue;
        };

        match &first[0..1] {
            "c" => {
                rest.split(',').for_each(|component| {
                    let mut component = component.split_whitespace();
                    let Some(name) = component.next() else {
                        return;
                    };
                    let size = match component.next().map(str::parse) {
                        Some(Ok(size)) => size,
                        _ => 0,
                    };
                    // Register our new component to the world with a layout specified by it's size
                    // SAFETY: [u64] is Send + Sync
                    let id = world.register_component_with_descriptor(unsafe {
                        ComponentDescriptor::new_with_layout(
                            name.to_string(),
                            StorageType::Table,
                            Layout::array::<u64>(size).unwrap(),
                            None,
                        )
                    });
                    let Some(info) = world.components().get_info(id) else {
                        return;
                    };
                    component_names.insert(name.to_string(), id);
                    component_info.insert(id, info.clone());
                    println!("Component {} created with id: {:?}", name, id.index());
                });
            }
            "s" => {
                let mut to_insert_ids = Vec::new();
                let mut to_insert_data = Vec::new();
                rest.split(',').for_each(|component| {
                    let mut component = component.split_whitespace();
                    let Some(name) = component.next() else {
                        return;
                    };

                    // Get the id for the component with the given name
                    let Some(&id) = component_names.get(name) else {
                        println!("Component {name} does not exist");
                        return;
                    };

                    // Calculate the length for the array based on the layout created for this component id
                    let info = world.components().get_info(id).unwrap();
                    let len = info.layout().size() / size_of::<u64>();
                    let mut values: Vec<u64> = component
                        .take(len)
                        .filter_map(|value| value.parse::<u64>().ok())
                        .collect();
                    values.resize(len, 0);

                    // Collect the id and array to be inserted onto our entity
                    to_insert_ids.push(id);
                    to_insert_data.push(values);
                });

                let mut entity = world.spawn_empty();

                // Construct an `OwningPtr` for each component in `to_insert_data`
                let to_insert_ptr = to_owning_ptrs(&mut to_insert_data);

                // SAFETY:
                // - Component ids have been taken from the same world
                // - Each array is created to the layout specified in the world
                unsafe {
                    entity.insert_by_ids(&to_insert_ids, to_insert_ptr.into_iter());
                }

                println!("Entity spawned with id: {:?}", entity.id());
            }
            "q" => {
                let mut builder = QueryBuilder::<FilteredEntityMut>::new(&mut world);
                parse_query(rest, &mut builder, &component_names);
                let mut query = builder.build();

                query.iter_mut(&mut world).for_each(|filtered_entity| {
                    #[allow(deprecated)]
                    let terms = filtered_entity
                        .access()
                        .component_reads_and_writes()
                        .0
                        .map(|id| {
                            let ptr = filtered_entity.get_by_id(id).unwrap();
                            let info = component_info.get(&id).unwrap();
                            let len = info.layout().size() / size_of::<u64>();

                            // SAFETY:
                            // - All components are created with layout [u64]
                            // - len is calculated from the component descriptor
                            let data = unsafe {
                                std::slice::from_raw_parts_mut(
                                    ptr.assert_unique().as_ptr().cast::<u64>(),
                                    len,
                                )
                            };

                            // If we have write access, increment each value once
                            if filtered_entity.access().has_component_write(id) {
                                data.iter_mut().for_each(|data| {
                                    *data += 1;
                                });
                            }

                            format!("{}: {:?}", info.name(), data[0..len].to_vec())
                        })
                        .collect::<Vec<_>>()
                        .join(", ");

                    println!("{:?}: {}", filtered_entity.id(), terms);
                });
            }
            _ => continue,
        }
    }
}

pub fn spawn<B>(&mut self, bundle: B) -> EntityWorldMut<'_>

where B: Bundle,

Spawns a new Entity with a given Bundle of components and returns a corresponding EntityWorldMut, which can be used to add components to the entity or retrieve its id. In case large batches of entities need to be spawned, consider using World::spawn_batch instead.

use bevy_ecs::{bundle::Bundle, component::Component, world::World};

#[derive(Component)]
struct Position {
  x: f32,
  y: f32,
}

#[derive(Component)]
struct Velocity {
    x: f32,
    y: f32,
};

#[derive(Component)]
struct Name(&'static str);

#[derive(Bundle)]
struct PhysicsBundle {
    position: Position,
    velocity: Velocity,
}

let mut world = World::new();

// `spawn` can accept a single component:
world.spawn(Position { x: 0.0, y: 0.0 });

// It can also accept a tuple of components:
world.spawn((
    Position { x: 0.0, y: 0.0 },
    Velocity { x: 1.0, y: 1.0 },
));

// Or it can accept a pre-defined Bundle of components:
world.spawn(PhysicsBundle {
    position: Position { x: 2.0, y: 2.0 },
    velocity: Velocity { x: 0.0, y: 4.0 },
});

let entity = world
    // Tuples can also mix Bundles and Components
    .spawn((
        PhysicsBundle {
            position: Position { x: 2.0, y: 2.0 },
            velocity: Velocity { x: 0.0, y: 4.0 },
        },
        Name("Elaina Proctor"),
    ))
    // Calling id() will return the unique identifier for the spawned entity
    .id();
let position = world.entity(entity).get::<Position>().unwrap();
assert_eq!(position.x, 2.0);

Examples found in repository

examples/ecs/one_shot_systems.rs (line 52)

fn setup_with_world(world: &mut World) {
    // We can run it once manually
    world.run_system_once(system_b).unwrap();
    // Or with a Callback
    let system_id = world.register_system(system_b);
    world.spawn((Callback(system_id), B));
}

examples/animation/animated_fox.rs (lines 291-304)

fn spawn_particle<M: Material>(
    mesh: Handle<Mesh>,
    material: Handle<M>,
    translation: Vec3,
    lifetime: f32,
    size: f32,
    velocity: Vec3,
) -> impl Command {
    move |world: &mut World| {
        world.spawn((
            Particle {
                lifeteime_timer: Timer::from_seconds(lifetime, TimerMode::Once),
                size,
                velocity,
            },
            Mesh3d(mesh),
            MeshMaterial3d(material),
            Transform {
                translation,
                scale: Vec3::splat(size),
                ..Default::default()
            },
        ));
    }
}

examples/ecs/ecs_guide.rs (lines 255-261)

fn exclusive_player_system(world: &mut World) {
    // this does the same thing as "new_player_system"
    let total_players = world.resource_mut::<GameState>().total_players;
    let should_add_player = {
        let game_rules = world.resource::<GameRules>();
        let add_new_player = random::<bool>();
        add_new_player && total_players < game_rules.max_players
    };
    // Randomly add a new player
    if should_add_player {
        println!("Player {} has joined the game!", total_players + 1);
        world.spawn((
            Player {
                name: format!("Player {}", total_players + 1),
            },
            Score { value: 0 },
            PlayerStreak::None,
        ));

        let mut game_state = world.resource_mut::<GameState>();
        game_state.total_players += 1;
    }
}

examples/scene/scene.rs (lines 110-115)

fn save_scene_system(world: &mut World) {
    // Scenes can be created from any ECS World.
    // You can either create a new one for the scene or use the current World.
    // For demonstration purposes, we'll create a new one.
    let mut scene_world = World::new();

    // The `TypeRegistry` resource contains information about all registered types (including components).
    // This is used to construct scenes, so we'll want to ensure that our previous type registrations
    // exist in this new scene world as well.
    // To do this, we can simply clone the `AppTypeRegistry` resource.
    let type_registry = world.resource::<AppTypeRegistry>().clone();
    scene_world.insert_resource(type_registry);

    let mut component_b = ComponentB::from_world(world);
    component_b.value = "hello".to_string();
    scene_world.spawn((
        component_b,
        ComponentA { x: 1.0, y: 2.0 },
        Transform::IDENTITY,
        Name::new("joe"),
    ));
    scene_world.spawn(ComponentA { x: 3.0, y: 4.0 });
    scene_world.insert_resource(ResourceA { score: 1 });

    // With our sample world ready to go, we can now create our scene using DynamicScene or DynamicSceneBuilder.
    // For simplicity, we will create our scene using DynamicScene:
    let scene = DynamicScene::from_world(&scene_world);

    // Scenes can be serialized like this:
    let type_registry = world.resource::<AppTypeRegistry>();
    let type_registry = type_registry.read();
    let serialized_scene = scene.serialize(&type_registry).unwrap();

    // Showing the scene in the console
    info!("{}", serialized_scene);

    // Writing the scene to a new file. Using a task to avoid calling the filesystem APIs in a system
    // as they are blocking
    // This can't work in Wasm as there is no filesystem access
    #[cfg(not(target_arch = "wasm32"))]
    IoTaskPool::get()
        .spawn(async move {
            // Write the scene RON data to file
            File::create(format!("assets/{NEW_SCENE_FILE_PATH}"))
                .and_then(|mut file| file.write(serialized_scene.as_bytes()))
                .expect("Error while writing scene to file");
        })
        .detach();
}

pub fn spawn_batch<I>( &mut self, iter: I, ) -> SpawnBatchIter<'_, <I as IntoIterator>::IntoIter>

where I: IntoIterator, <I as IntoIterator>::Item: Bundle,

Spawns a batch of entities with the same component Bundle type. Takes a given Bundle iterator and returns a corresponding Entity iterator. This is more efficient than spawning entities and adding components to them individually using World::spawn, but it is limited to spawning entities with the same Bundle type, whereas spawning individually is more flexible.

use bevy_ecs::{component::Component, entity::Entity, world::World};

#[derive(Component)]
struct Str(&'static str);
#[derive(Component)]
struct Num(u32);

let mut world = World::new();
let entities = world.spawn_batch(vec![
  (Str("a"), Num(0)), // the first entity
  (Str("b"), Num(1)), // the second entity
]).collect::<Vec<Entity>>();

assert_eq!(entities.len(), 2);

pub fn get<T>(&self, entity: Entity) -> Option<&T>

where T: Component,

Retrieves a reference to the given entity’s Component of the given type. Returns None if the entity does not have a Component of the given type.

use bevy_ecs::{component::Component, world::World};

#[derive(Component)]
struct Position {
  x: f32,
  y: f32,
}

let mut world = World::new();
let entity = world.spawn(Position { x: 0.0, y: 0.0 }).id();
let position = world.get::<Position>(entity).unwrap();
assert_eq!(position.x, 0.0);

Examples found in repository

examples/ecs/component_hooks.rs (line 73)

fn setup(world: &mut World) {
    // In order to register component hooks the component must:
    // - not be currently in use by any entities in the world
    // - not already have a hook of that kind registered
    // This is to prevent overriding hooks defined in plugins and other crates as well as keeping things fast
    world
        .register_component_hooks::<MyComponent>()
        // There are 4 component lifecycle hooks: `on_add`, `on_insert`, `on_replace` and `on_remove`
        // A hook has 3 arguments:
        // - a `DeferredWorld`, this allows access to resource and component data as well as `Commands`
        // - the entity that triggered the hook
        // - the component id of the triggering component, this is mostly used for dynamic components
        //
        // `on_add` will trigger when a component is inserted onto an entity without it
        .on_add(|mut world, entity, component_id| {
            // You can access component data from within the hook
            let value = world.get::<MyComponent>(entity).unwrap().0;
            println!("Component: {component_id:?} added to: {entity:?} with value {value:?}");
            // Or access resources
            world
                .resource_mut::<MyComponentIndex>()
                .insert(value, entity);
            // Or send events
            world.send_event(MyEvent);
        })
        // `on_insert` will trigger when a component is inserted onto an entity,
        // regardless of whether or not it already had it and after `on_add` if it ran
        .on_insert(|world, _, _| {
            println!("Current Index: {:?}", world.resource::<MyComponentIndex>());
        })
        // `on_replace` will trigger when a component is inserted onto an entity that already had it,
        // and runs before the value is replaced.
        // Also triggers when a component is removed from an entity, and runs before `on_remove`
        .on_replace(|mut world, entity, _| {
            let value = world.get::<MyComponent>(entity).unwrap().0;
            world.resource_mut::<MyComponentIndex>().remove(&value);
        })
        // `on_remove` will trigger when a component is removed from an entity,
        // since it runs before the component is removed you can still access the component data
        .on_remove(|mut world, entity, component_id| {
            let value = world.get::<MyComponent>(entity).unwrap().0;
            println!("Component: {component_id:?} removed from: {entity:?} with value {value:?}");
            // You can also issue commands through `.commands()`
            world.commands().entity(entity).despawn();
        });
}

pub fn get_mut<T>(&mut self, entity: Entity) -> Option<Mut<'_, T>>

where T: Component,

Retrieves a mutable reference to the given entity’s Component of the given type. Returns None if the entity does not have a Component of the given type.

use bevy_ecs::{component::Component, world::World};

#[derive(Component)]
struct Position {
  x: f32,
  y: f32,
}

let mut world = World::new();
let entity = world.spawn(Position { x: 0.0, y: 0.0 }).id();
let mut position = world.get_mut::<Position>(entity).unwrap();
position.x = 1.0;

pub fn despawn(&mut self, entity: Entity) -> bool

Despawns the given entity, if it exists. This will also remove all of the entity’s Components. Returns true if the entity is successfully despawned and false if the entity does not exist.

Note

This won’t clean up external references to the entity (such as parent-child relationships if you’re using bevy_hierarchy), which may leave the world in an invalid state.

use bevy_ecs::{component::Component, world::World};

#[derive(Component)]
struct Position {
  x: f32,
  y: f32,
}

let mut world = World::new();
let entity = world.spawn(Position { x: 0.0, y: 0.0 }).id();
assert!(world.despawn(entity));
assert!(world.get_entity(entity).is_err());
assert!(world.get::<Position>(entity).is_none());

pub fn try_despawn(&mut self, entity: Entity) -> bool

Performs the same function as Self::despawn but does not emit a warning if the entity does not exist.

pub fn clear_trackers(&mut self)

Clears the internal component tracker state.

The world maintains some internal state about changed and removed components. This state is used by RemovedComponents to provide access to the entities that had a specific type of component removed since last tick.

The state is also used for change detection when accessing components and resources outside of a system, for example via World::get_mut() or World::get_resource_mut().

By clearing this internal state, the world “forgets” about those changes, allowing a new round of detection to be recorded.

When using bevy_ecs as part of the full Bevy engine, this method is called automatically by bevy_app::App::update and bevy_app::SubApp::update, so you don’t need to call it manually. When using bevy_ecs as a separate standalone crate however, you do need to call this manually.

// a whole new world
let mut world = World::new();

// you changed it
let entity = world.spawn(Transform::default()).id();

// change is detected
let transform = world.get_mut::<Transform>(entity).unwrap();
assert!(transform.is_changed());

// update the last change tick
world.clear_trackers();

// change is no longer detected
let transform = world.get_mut::<Transform>(entity).unwrap();
assert!(!transform.is_changed());

pub fn query<D>(&mut self) -> QueryState<D>

where D: QueryData,

Returns QueryState for the given QueryData, which is used to efficiently run queries on the World by storing and reusing the QueryState.

use bevy_ecs::{component::Component, entity::Entity, world::World};

#[derive(Component, Debug, PartialEq)]
struct Position {
  x: f32,
  y: f32,
}

#[derive(Component)]
struct Velocity {
  x: f32,
  y: f32,
}

let mut world = World::new();
let entities = world.spawn_batch(vec![
    (Position { x: 0.0, y: 0.0}, Velocity { x: 1.0, y: 0.0 }),
    (Position { x: 0.0, y: 0.0}, Velocity { x: 0.0, y: 1.0 }),
]).collect::<Vec<Entity>>();

let mut query = world.query::<(&mut Position, &Velocity)>();
for (mut position, velocity) in query.iter_mut(&mut world) {
   position.x += velocity.x;
   position.y += velocity.y;
}

assert_eq!(world.get::<Position>(entities[0]).unwrap(), &Position { x: 1.0, y: 0.0 });
assert_eq!(world.get::<Position>(entities[1]).unwrap(), &Position { x: 0.0, y: 1.0 });

To iterate over entities in a deterministic order, sort the results of the query using the desired component as a key. Note that this requires fetching the whole result set from the query and allocation of a Vec to store it.

use bevy_ecs::{component::Component, entity::Entity, world::World};

#[derive(Component, PartialEq, Eq, PartialOrd, Ord, Debug)]
struct Order(i32);
#[derive(Component, PartialEq, Debug)]
struct Label(&'static str);

let mut world = World::new();
let a = world.spawn((Order(2), Label("second"))).id();
let b = world.spawn((Order(3), Label("third"))).id();
let c = world.spawn((Order(1), Label("first"))).id();
let mut entities = world.query::<(Entity, &Order, &Label)>()
    .iter(&world)
    .collect::<Vec<_>>();
// Sort the query results by their `Order` component before comparing
// to expected results. Query iteration order should not be relied on.
entities.sort_by_key(|e| e.1);
assert_eq!(entities, vec![
    (c, &Order(1), &Label("first")),
    (a, &Order(2), &Label("second")),
    (b, &Order(3), &Label("third")),
]);

pub fn query_filtered<D, F>(&mut self) -> QueryState<D, F>

where D: QueryData, F: QueryFilter,

Returns QueryState for the given filtered QueryData, which is used to efficiently run queries on the World by storing and reusing the QueryState.

use bevy_ecs::{component::Component, entity::Entity, world::World, query::With};

#[derive(Component)]
struct A;
#[derive(Component)]
struct B;

let mut world = World::new();
let e1 = world.spawn(A).id();
let e2 = world.spawn((A, B)).id();

let mut query = world.query_filtered::<Entity, With<B>>();
let matching_entities = query.iter(&world).collect::<Vec<Entity>>();

assert_eq!(matching_entities, vec![e2]);

pub fn removed<T>(&self) -> impl Iterator<Item = Entity>

where T: Component,

Returns an iterator of entities that had components of type T removed since the last call to World::clear_trackers.

pub fn removed_with_id( &self, component_id: ComponentId, ) -> impl Iterator<Item = Entity>

Returns an iterator of entities that had components with the given component_id removed since the last call to World::clear_trackers.

pub fn register_resource_with_descriptor( &mut self, descriptor: ComponentDescriptor, ) -> ComponentId

Registers a new Resource type and returns the ComponentId created for it.

This enables the dynamic registration of new Resource definitions at runtime for advanced use cases.

Note

Registering a Resource does not insert it into World. For insertion, you could use World::insert_resource_by_id.

pub fn init_resource<R>(&mut self) -> ComponentId

where R: Resource + FromWorld,

Initializes a new resource and returns the ComponentId created for it.

If the resource already exists, nothing happens.

The value given by the FromWorld::from_world method will be used. Note that any resource with the Default trait automatically implements FromWorld, and those default values will be here instead.

pub fn insert_resource<R>(&mut self, value: R)

where R: Resource,

Inserts a new resource with the given value.

Resources are “unique” data of a given type. If you insert a resource of a type that already exists, you will overwrite any existing data.

Examples found in repository

examples/scene/scene.rs (line 106)

fn save_scene_system(world: &mut World) {
    // Scenes can be created from any ECS World.
    // You can either create a new one for the scene or use the current World.
    // For demonstration purposes, we'll create a new one.
    let mut scene_world = World::new();

    // The `TypeRegistry` resource contains information about all registered types (including components).
    // This is used to construct scenes, so we'll want to ensure that our previous type registrations
    // exist in this new scene world as well.
    // To do this, we can simply clone the `AppTypeRegistry` resource.
    let type_registry = world.resource::<AppTypeRegistry>().clone();
    scene_world.insert_resource(type_registry);

    let mut component_b = ComponentB::from_world(world);
    component_b.value = "hello".to_string();
    scene_world.spawn((
        component_b,
        ComponentA { x: 1.0, y: 2.0 },
        Transform::IDENTITY,
        Name::new("joe"),
    ));
    scene_world.spawn(ComponentA { x: 3.0, y: 4.0 });
    scene_world.insert_resource(ResourceA { score: 1 });

    // With our sample world ready to go, we can now create our scene using DynamicScene or DynamicSceneBuilder.
    // For simplicity, we will create our scene using DynamicScene:
    let scene = DynamicScene::from_world(&scene_world);

    // Scenes can be serialized like this:
    let type_registry = world.resource::<AppTypeRegistry>();
    let type_registry = type_registry.read();
    let serialized_scene = scene.serialize(&type_registry).unwrap();

    // Showing the scene in the console
    info!("{}", serialized_scene);

    // Writing the scene to a new file. Using a task to avoid calling the filesystem APIs in a system
    // as they are blocking
    // This can't work in Wasm as there is no filesystem access
    #[cfg(not(target_arch = "wasm32"))]
    IoTaskPool::get()
        .spawn(async move {
            // Write the scene RON data to file
            File::create(format!("assets/{NEW_SCENE_FILE_PATH}"))
                .and_then(|mut file| file.write(serialized_scene.as_bytes()))
                .expect("Error while writing scene to file");
        })
        .detach();
}

pub fn init_non_send_resource<R>(&mut self) -> ComponentId

where R: 'static + FromWorld,

Initializes a new non-send resource and returns the ComponentId created for it.

If the resource already exists, nothing happens.

The value given by the FromWorld::from_world method will be used. Note that any resource with the Default trait automatically implements FromWorld, and those default values will be here instead.

Panics

Panics if called from a thread other than the main thread.

pub fn insert_non_send_resource<R>(&mut self, value: R)

where R: 'static,

Inserts a new non-send resource with the given value.

NonSend resources cannot be sent across threads, and do not need the Send + Sync bounds. Systems with NonSend resources are always scheduled on the main thread.

Panics

If a value is already present, this function will panic if called from a different thread than where the original value was inserted from.

pub fn remove_resource<R>(&mut self) -> Option<R>

where R: Resource,

Removes the resource of a given type and returns it, if it exists. Otherwise returns None.

pub fn remove_non_send_resource<R>(&mut self) -> Option<R>

where R: 'static,

Removes a !Send resource from the world and returns it, if present.

NonSend resources cannot be sent across threads, and do not need the Send + Sync bounds. Systems with NonSend resources are always scheduled on the main thread.

Returns None if a value was not previously present.

Panics

If a value is present, this function will panic if called from a different thread than where the value was inserted from.

pub fn contains_resource<R>(&self) -> bool

where R: Resource,

Returns true if a resource of type R exists. Otherwise returns false.

pub fn contains_resource_by_id(&self, component_id: ComponentId) -> bool

Returns true if a resource with provided component_id exists. Otherwise returns false.

pub fn contains_non_send<R>(&self) -> bool

where R: 'static,

Returns true if a resource of type R exists. Otherwise returns false.

pub fn contains_non_send_by_id(&self, component_id: ComponentId) -> bool

Returns true if a resource with provided component_id exists. Otherwise returns false.

pub fn is_resource_added<R>(&self) -> bool

where R: Resource,

Returns true if a resource of type R exists and was added since the world’s last_change_tick. Otherwise, this returns false.

This means that:

• When called from an exclusive system, this will check for additions since the system last ran.
• When called elsewhere, this will check for additions since the last time that World::clear_trackers was called.

pub fn is_resource_added_by_id(&self, component_id: ComponentId) -> bool

Returns true if a resource with id component_id exists and was added since the world’s last_change_tick. Otherwise, this returns false.

This means that:

• When called from an exclusive system, this will check for additions since the system last ran.
• When called elsewhere, this will check for additions since the last time that World::clear_trackers was called.

pub fn is_resource_changed<R>(&self) -> bool

where R: Resource,

Returns true if a resource of type R exists and was modified since the world’s last_change_tick. Otherwise, this returns false.

This means that:

• When called from an exclusive system, this will check for changes since the system last ran.
• When called elsewhere, this will check for changes since the last time that World::clear_trackers was called.

pub fn is_resource_changed_by_id(&self, component_id: ComponentId) -> bool

Returns true if a resource with id component_id exists and was modified since the world’s last_change_tick. Otherwise, this returns false.

This means that:

• When called from an exclusive system, this will check for changes since the system last ran.
• When called elsewhere, this will check for changes since the last time that World::clear_trackers was called.

pub fn get_resource_change_ticks<R>(&self) -> Option<ComponentTicks>

where R: Resource,

Retrieves the change ticks for the given resource.

pub fn get_resource_change_ticks_by_id( &self, component_id: ComponentId, ) -> Option<ComponentTicks>

Retrieves the change ticks for the given ComponentId.

You should prefer to use the typed API World::get_resource_change_ticks where possible.

pub fn resource<R>(&self) -> &R

where R: Resource,

Gets a reference to the resource of the given type

Panics

Panics if the resource does not exist. Use get_resource instead if you want to handle this case.

If you want to instead insert a value if the resource does not exist, use get_resource_or_insert_with.

Examples found in repository

examples/scene/scene.rs (line 46)

    fn from_world(world: &mut World) -> Self {
        let time = world.resource::<Time>();
        ComponentB {
            _time_since_startup: time.elapsed(),
            value: "Default Value".to_string(),
        }
    }
}

// Resources can be serialized in scenes as well, with the same requirements `Component`s have.
#[derive(Resource, Reflect, Default)]
#[reflect(Resource)]
struct ResourceA {
    pub score: u32,
}

// The initial scene file will be loaded below and not change when the scene is saved
const SCENE_FILE_PATH: &str = "scenes/load_scene_example.scn.ron";

// The new, updated scene data will be saved here so that you can see the changes
const NEW_SCENE_FILE_PATH: &str = "scenes/load_scene_example-new.scn.ron";

fn load_scene_system(mut commands: Commands, asset_server: Res<AssetServer>) {
    // Spawning a DynamicSceneRoot creates a new entity and spawns new instances
    // of the given scene's entities as children of that entity.
    // Scenes can be loaded just like any other asset.
    commands.spawn(DynamicSceneRoot(asset_server.load(SCENE_FILE_PATH)));
}

// This system logs all ComponentA components in our world. Try making a change to a ComponentA in
// load_scene_example.scn. If you enable the `file_watcher` cargo feature you should immediately see
// the changes appear in the console whenever you make a change.
fn log_system(
    query: Query<(Entity, &ComponentA), Changed<ComponentA>>,
    res: Option<Res<ResourceA>>,
) {
    for (entity, component_a) in &query {
        info!("  Entity({})", entity.index());
        info!(
            "    ComponentA: {{ x: {} y: {} }}\n",
            component_a.x, component_a.y
        );
    }
    if let Some(res) = res {
        if res.is_added() {
            info!("  New ResourceA: {{ score: {} }}\n", res.score);
        }
    }
}

fn save_scene_system(world: &mut World) {
    // Scenes can be created from any ECS World.
    // You can either create a new one for the scene or use the current World.
    // For demonstration purposes, we'll create a new one.
    let mut scene_world = World::new();

    // The `TypeRegistry` resource contains information about all registered types (including components).
    // This is used to construct scenes, so we'll want to ensure that our previous type registrations
    // exist in this new scene world as well.
    // To do this, we can simply clone the `AppTypeRegistry` resource.
    let type_registry = world.resource::<AppTypeRegistry>().clone();
    scene_world.insert_resource(type_registry);

    let mut component_b = ComponentB::from_world(world);
    component_b.value = "hello".to_string();
    scene_world.spawn((
        component_b,
        ComponentA { x: 1.0, y: 2.0 },
        Transform::IDENTITY,
        Name::new("joe"),
    ));
    scene_world.spawn(ComponentA { x: 3.0, y: 4.0 });
    scene_world.insert_resource(ResourceA { score: 1 });

    // With our sample world ready to go, we can now create our scene using DynamicScene or DynamicSceneBuilder.
    // For simplicity, we will create our scene using DynamicScene:
    let scene = DynamicScene::from_world(&scene_world);

    // Scenes can be serialized like this:
    let type_registry = world.resource::<AppTypeRegistry>();
    let type_registry = type_registry.read();
    let serialized_scene = scene.serialize(&type_registry).unwrap();

    // Showing the scene in the console
    info!("{}", serialized_scene);

    // Writing the scene to a new file. Using a task to avoid calling the filesystem APIs in a system
    // as they are blocking
    // This can't work in Wasm as there is no filesystem access
    #[cfg(not(target_arch = "wasm32"))]
    IoTaskPool::get()
        .spawn(async move {
            // Write the scene RON data to file
            File::create(format!("assets/{NEW_SCENE_FILE_PATH}"))
                .and_then(|mut file| file.write(serialized_scene.as_bytes()))
                .expect("Error while writing scene to file");
        })
        .detach();
}

examples/shader/custom_shader_instancing.rs (line 202)

    fn from_world(world: &mut World) -> Self {
        let mesh_pipeline = world.resource::<MeshPipeline>();

        CustomPipeline {
            shader: world.load_asset(SHADER_ASSET_PATH),
            mesh_pipeline: mesh_pipeline.clone(),
        }
    }

examples/shader/specialized_mesh_pipeline.rs (line 172)

    fn from_world(world: &mut World) -> Self {
        // Load the shader
        let shader_handle: Handle<Shader> = world.resource::<AssetServer>().load(SHADER_ASSET_PATH);
        Self {
            mesh_pipeline: MeshPipeline::from_world(world),
            shader_handle,
        }
    }

tests/ecs/ambiguity_detection.rs (line 73)

fn count_ambiguities(sub_app: &SubApp) -> AmbiguitiesCount {
    let schedules = sub_app.world().resource::<Schedules>();
    let mut ambiguities = HashMap::new();
    for (_, schedule) in schedules.iter() {
        let ambiguities_in_schedule = schedule.graph().conflicting_systems().len();
        ambiguities.insert(schedule.label(), ambiguities_in_schedule);
    }
    AmbiguitiesCount(ambiguities)
}

examples/shader/custom_phase_item.rs (line 352)

    fn from_world(world: &mut World) -> Self {
        let render_device = world.resource::<RenderDevice>();
        let render_queue = world.resource::<RenderQueue>();

        // Create the vertex and index buffers.
        let mut vbo = RawBufferVec::new(BufferUsages::VERTEX);
        let mut ibo = RawBufferVec::new(BufferUsages::INDEX);

        for vertex in &VERTICES {
            vbo.push(*vertex);
        }
        for index in 0..3 {
            ibo.push(index);
        }

        // These two lines are required in order to trigger the upload to GPU.
        vbo.write_buffer(render_device, render_queue);
        ibo.write_buffer(render_device, render_queue);

        CustomPhaseItemBuffers {
            vertices: vbo,
            indices: ibo,
        }
    }
}

impl FromWorld for CustomPhasePipeline {
    fn from_world(world: &mut World) -> Self {
        // Load and compile the shader in the background.
        let asset_server = world.resource::<AssetServer>();

        CustomPhasePipeline {
            shader: asset_server.load("shaders/custom_phase_item.wgsl"),
        }
    }

examples/shader/texture_binding_array.rs (line 50)

    fn finish(&self, app: &mut App) {
        let Some(render_app) = app.get_sub_app_mut(RenderApp) else {
            return;
        };

        let render_device = render_app.world().resource::<RenderDevice>();

        // Check if the device support the required feature. If not, exit the example.
        // In a real application, you should setup a fallback for the missing feature
        if !render_device
            .features()
            .contains(WgpuFeatures::SAMPLED_TEXTURE_AND_STORAGE_BUFFER_ARRAY_NON_UNIFORM_INDEXING)
        {
            error!(
                "Render device doesn't support feature \
SAMPLED_TEXTURE_AND_STORAGE_BUFFER_ARRAY_NON_UNIFORM_INDEXING, \
which is required for texture binding arrays"
            );
            exit(1);
        }
    }

Additional examples can be found in:

• examples/ecs/ecs_guide.rs
• examples/shader/gpu_readback.rs
• examples/shader/compute_shader_game_of_life.rs
• examples/ecs/component_hooks.rs
• examples/shader/custom_post_processing.rs

pub fn resource_ref<R>(&self) -> Ref<'_, R>

where R: Resource,

Gets a reference to the resource of the given type

Panics

Panics if the resource does not exist. Use get_resource_ref instead if you want to handle this case.

If you want to instead insert a value if the resource does not exist, use get_resource_or_insert_with.

pub fn resource_mut<R>(&mut self) -> Mut<'_, R>

where R: Resource,

Gets a mutable reference to the resource of the given type

Panics

Panics if the resource does not exist. Use get_resource_mut instead if you want to handle this case.

If you want to instead insert a value if the resource does not exist, use get_resource_or_insert_with.

Examples found in repository

examples/animation/animated_fox.rs (line 324)

    fn from_world(world: &mut World) -> Self {
        Self {
            mesh: world.resource_mut::<Assets<Mesh>>().add(Sphere::new(10.0)),
            material: world
                .resource_mut::<Assets<StandardMaterial>>()
                .add(StandardMaterial {
                    base_color: WHITE.into(),
                    ..Default::default()
                }),
        }
    }

tests/ecs/ambiguity_detection.rs (line 60)

fn configure_ambiguity_detection(sub_app: &mut SubApp) {
    let mut schedules = sub_app.world_mut().resource_mut::<Schedules>();
    for (_, schedule) in schedules.iter_mut() {
        schedule.set_build_settings(ScheduleBuildSettings {
            // NOTE: you can change this to `LogLevel::Ignore` to easily see the current number of ambiguities.
            ambiguity_detection: LogLevel::Warn,
            use_shortnames: false,
            ..default()
        });
    }
}

examples/app/custom_loop.rs (line 19)

fn my_runner(mut app: App) -> AppExit {
    // Finalize plugin building, including running any necessary clean-up.
    // This is normally completed by the default runner.
    app.finish();
    app.cleanup();

    println!("Type stuff into the console");
    for line in io::stdin().lines() {
        {
            let mut input = app.world_mut().resource_mut::<Input>();
            input.0 = line.unwrap();
        }
        app.update();

        if let Some(exit) = app.should_exit() {
            return exit;
        }
    }

    AppExit::Success
}

examples/shader/gpu_readback.rs (line 60)

    fn finish(&self, app: &mut App) {
        let render_app = app.sub_app_mut(RenderApp);
        render_app.init_resource::<ComputePipeline>().add_systems(
            Render,
            prepare_bind_group
                .in_set(RenderSet::PrepareBindGroups)
                // We don't need to recreate the bind group every frame
                .run_if(not(resource_exists::<GpuBufferBindGroup>)),
        );

        // Add the compute node as a top level node to the render graph
        // This means it will only execute once per frame
        render_app
            .world_mut()
            .resource_mut::<RenderGraph>()
            .add_node(ComputeNodeLabel, ComputeNode::default());
    }

examples/shader/compute_shader_game_of_life.rs (line 111)

    fn build(&self, app: &mut App) {
        // Extract the game of life image resource from the main world into the render world
        // for operation on by the compute shader and display on the sprite.
        app.add_plugins(ExtractResourcePlugin::<GameOfLifeImages>::default());
        let render_app = app.sub_app_mut(RenderApp);
        render_app.add_systems(
            Render,
            prepare_bind_group.in_set(RenderSet::PrepareBindGroups),
        );

        let mut render_graph = render_app.world_mut().resource_mut::<RenderGraph>();
        render_graph.add_node(GameOfLifeLabel, GameOfLifeNode::default());
        render_graph.add_node_edge(GameOfLifeLabel, bevy::render::graph::CameraDriverLabel);
    }

examples/ecs/ecs_guide.rs (line 246)

fn exclusive_player_system(world: &mut World) {
    // this does the same thing as "new_player_system"
    let total_players = world.resource_mut::<GameState>().total_players;
    let should_add_player = {
        let game_rules = world.resource::<GameRules>();
        let add_new_player = random::<bool>();
        add_new_player && total_players < game_rules.max_players
    };
    // Randomly add a new player
    if should_add_player {
        println!("Player {} has joined the game!", total_players + 1);
        world.spawn((
            Player {
                name: format!("Player {}", total_players + 1),
            },
            Score { value: 0 },
            PlayerStreak::None,
        ));

        let mut game_state = world.resource_mut::<GameState>();
        game_state.total_players += 1;
    }
}

Additional examples can be found in:

• examples/app/headless_renderer.rs
• examples/2d/mesh2d_manual.rs
• examples/games/stepping.rs
• examples/time/time.rs
• examples/ecs/custom_schedule.rs
• examples/shader/custom_post_processing.rs
• examples/ecs/system_stepping.rs

pub fn get_resource<R>(&self) -> Option<&R>

where R: Resource,

Gets a reference to the resource of the given type if it exists

Examples found in repository

examples/app/headless_renderer.rs (line 341)

    fn run(
        &self,
        _graph: &mut RenderGraphContext,
        render_context: &mut RenderContext,
        world: &World,
    ) -> Result<(), NodeRunError> {
        let image_copiers = world.get_resource::<ImageCopiers>().unwrap();
        let gpu_images = world
            .get_resource::<RenderAssets<bevy::render::texture::GpuImage>>()
            .unwrap();

        for image_copier in image_copiers.iter() {
            if !image_copier.enabled() {
                continue;
            }

            let src_image = gpu_images.get(&image_copier.src_image).unwrap();

            let mut encoder = render_context
                .render_device()
                .create_command_encoder(&CommandEncoderDescriptor::default());

            let block_dimensions = src_image.texture_format.block_dimensions();
            let block_size = src_image.texture_format.block_copy_size(None).unwrap();

            // Calculating correct size of image row because
            // copy_texture_to_buffer can copy image only by rows aligned wgpu::COPY_BYTES_PER_ROW_ALIGNMENT
            // That's why image in buffer can be little bit wider
            // This should be taken into account at copy from buffer stage
            let padded_bytes_per_row = RenderDevice::align_copy_bytes_per_row(
                (src_image.size.x as usize / block_dimensions.0 as usize) * block_size as usize,
            );

            let texture_extent = Extent3d {
                width: src_image.size.x,
                height: src_image.size.y,
                depth_or_array_layers: 1,
            };

            encoder.copy_texture_to_buffer(
                src_image.texture.as_image_copy(),
                ImageCopyBuffer {
                    buffer: &image_copier.buffer,
                    layout: ImageDataLayout {
                        offset: 0,
                        bytes_per_row: Some(
                            std::num::NonZero::<u32>::new(padded_bytes_per_row as u32)
                                .unwrap()
                                .into(),
                        ),
                        rows_per_image: None,
                    },
                },
                texture_extent,
            );

            let render_queue = world.get_resource::<RenderQueue>().unwrap();
            render_queue.submit(std::iter::once(encoder.finish()));
        }

        Ok(())
    }

pub fn get_resource_ref<R>(&self) -> Option<Ref<'_, R>>

where R: Resource,

Gets a reference including change detection to the resource of the given type if it exists.

pub fn get_resource_mut<R>(&mut self) -> Option<Mut<'_, R>>

where R: Resource,

Gets a mutable reference to the resource of the given type if it exists

Examples found in repository

examples/games/loading_screen.rs (line 310)

    fn update_pipelines_ready(mut main_world: ResMut<MainWorld>, pipelines: Res<PipelineCache>) {
        if let Some(mut pipelines_ready) = main_world.get_resource_mut::<PipelinesReady>() {
            pipelines_ready.0 = pipelines.waiting_pipelines().count() == 0;
        }
    }

pub fn get_resource_or_insert_with<R>( &mut self, func: impl FnOnce() -> R, ) -> Mut<'_, R>

where R: Resource,

Gets a mutable reference to the resource of type T if it exists, otherwise inserts the resource using the result of calling func.

Example

#[derive(Resource)]
struct MyResource(i32);

let my_res = world.get_resource_or_insert_with(|| MyResource(10));
assert_eq!(my_res.0, 10);

pub fn get_resource_or_init<R>(&mut self) -> Mut<'_, R>

where R: Resource + FromWorld,

Gets a mutable reference to the resource of type T if it exists, otherwise initializes the resource by calling its FromWorld implementation.

Example

#[derive(Resource)]
struct Foo(i32);

impl Default for Foo {
    fn default() -> Self {
        Self(15)
    }
}

#[derive(Resource)]
struct MyResource(i32);

impl FromWorld for MyResource {
    fn from_world(world: &mut World) -> Self {
        let foo = world.get_resource_or_init::<Foo>();
        Self(foo.0 * 2)
    }
}

let my_res = world.get_resource_or_init::<MyResource>();
assert_eq!(my_res.0, 30);

pub fn non_send_resource<R>(&self) -> &R

where R: 'static,

Gets an immutable reference to the non-send resource of the given type, if it exists.

Panics

Panics if the resource does not exist. Use get_non_send_resource instead if you want to handle this case.

This function will panic if it isn’t called from the same thread that the resource was inserted from.

pub fn non_send_resource_mut<R>(&mut self) -> Mut<'_, R>

where R: 'static,

Gets a mutable reference to the non-send resource of the given type, if it exists.

Panics

Panics if the resource does not exist. Use get_non_send_resource_mut instead if you want to handle this case.

This function will panic if it isn’t called from the same thread that the resource was inserted from.

pub fn get_non_send_resource<R>(&self) -> Option<&R>

where R: 'static,

Gets a reference to the non-send resource of the given type, if it exists. Otherwise returns None.

Panics

This function will panic if it isn’t called from the same thread that the resource was inserted from.

pub fn get_non_send_resource_mut<R>(&mut self) -> Option<Mut<'_, R>>

where R: 'static,

Gets a mutable reference to the non-send resource of the given type, if it exists. Otherwise returns None.

Panics

This function will panic if it isn’t called from the same thread that the resource was inserted from.

pub fn insert_or_spawn_batch<I, B>( &mut self, iter: I, ) -> Result<(), Vec<Entity>>

where I: IntoIterator, <I as IntoIterator>::IntoIter: Iterator<Item = (Entity, B)>, B: Bundle,

For a given batch of (Entity, Bundle) pairs, either spawns each Entity with the given bundle (if the entity does not exist), or inserts the Bundle (if the entity already exists). This is faster than doing equivalent operations one-by-one. Returns Ok if all entities were successfully inserted into or spawned. Otherwise it returns an Err with a list of entities that could not be spawned or inserted into. A “spawn or insert” operation can only fail if an Entity is passed in with an “invalid generation” that conflicts with an existing Entity.

Note

Spawning a specific entity value is rarely the right choice. Most apps should use World::spawn_batch. This method should generally only be used for sharing entities across apps, and only when they have a scheme worked out to share an ID space (which doesn’t happen by default).

use bevy_ecs::{entity::Entity, world::World, component::Component};
#[derive(Component)]
struct A(&'static str);
#[derive(Component, PartialEq, Debug)]
struct B(f32);

let mut world = World::new();
let e0 = world.spawn_empty().id();
let e1 = world.spawn_empty().id();
world.insert_or_spawn_batch(vec![
  (e0, (A("a"), B(0.0))), // the first entity
  (e1, (A("b"), B(1.0))), // the second entity
]);

assert_eq!(world.get::<B>(e0), Some(&B(0.0)));

pub fn insert_batch<I, B>(&mut self, batch: I)

where I: IntoIterator, <I as IntoIterator>::IntoIter: Iterator<Item = (Entity, B)>, B: Bundle,

For a given batch of (Entity, Bundle) pairs, adds the Bundle of components to each Entity. This is faster than doing equivalent operations one-by-one.

A batch can be any type that implements IntoIterator containing (Entity, Bundle) tuples, such as a [Vec<(Entity, Bundle)>] or an array [(Entity, Bundle); N].

This will overwrite any previous values of components shared by the Bundle. See World::insert_batch_if_new to keep the old values instead.

Panics

This function will panic if any of the associated entities do not exist.

For the non-panicking version, see World::try_insert_batch.

pub fn insert_batch_if_new<I, B>(&mut self, batch: I)

where I: IntoIterator, <I as IntoIterator>::IntoIter: Iterator<Item = (Entity, B)>, B: Bundle,

For a given batch of (Entity, Bundle) pairs, adds the Bundle of components to each Entity without overwriting. This is faster than doing equivalent operations one-by-one.

A batch can be any type that implements IntoIterator containing (Entity, Bundle) tuples, such as a [Vec<(Entity, Bundle)>] or an array [(Entity, Bundle); N].

This is the same as World::insert_batch, but in case of duplicate components it will leave the old values instead of replacing them with new ones.

Panics

This function will panic if any of the associated entities do not exist.

For the non-panicking version, see World::try_insert_batch_if_new.

pub fn try_insert_batch<I, B>(&mut self, batch: I)

where I: IntoIterator, <I as IntoIterator>::IntoIter: Iterator<Item = (Entity, B)>, B: Bundle,

For a given batch of (Entity, Bundle) pairs, adds the Bundle of components to each Entity. This is faster than doing equivalent operations one-by-one.

A batch can be any type that implements IntoIterator containing (Entity, Bundle) tuples, such as a [Vec<(Entity, Bundle)>] or an array [(Entity, Bundle); N].

This will overwrite any previous values of components shared by the Bundle. See World::try_insert_batch_if_new to keep the old values instead.

This function silently fails by ignoring any entities that do not exist.

For the panicking version, see World::insert_batch.

pub fn try_insert_batch_if_new<I, B>(&mut self, batch: I)

where I: IntoIterator, <I as IntoIterator>::IntoIter: Iterator<Item = (Entity, B)>, B: Bundle,

For a given batch of (Entity, Bundle) pairs, adds the Bundle of components to each Entity without overwriting. This is faster than doing equivalent operations one-by-one.

A batch can be any type that implements IntoIterator containing (Entity, Bundle) tuples, such as a [Vec<(Entity, Bundle)>] or an array [(Entity, Bundle); N].

This is the same as World::try_insert_batch, but in case of duplicate components it will leave the old values instead of replacing them with new ones.

This function silently fails by ignoring any entities that do not exist.

For the panicking version, see World::insert_batch_if_new.

pub fn resource_scope<R, U>( &mut self, f: impl FnOnce(&mut World, Mut<'_, R>) -> U, ) -> U

where R: Resource,

Temporarily removes the requested resource from this World, runs custom user code, then re-adds the resource before returning.

This enables safe simultaneous mutable access to both a resource and the rest of the World. For more complex access patterns, consider using SystemState.

Example

use bevy_ecs::prelude::*;
#[derive(Resource)]
struct A(u32);
#[derive(Component)]
struct B(u32);
let mut world = World::new();
world.insert_resource(A(1));
let entity = world.spawn(B(1)).id();

world.resource_scope(|world, mut a: Mut<A>| {
    let b = world.get_mut::<B>(entity).unwrap();
    a.0 += b.0;
});
assert_eq!(world.get_resource::<A>().unwrap().0, 2);

pub fn send_event<E>(&mut self, event: E) -> Option<EventId<E>>

where E: Event,

Sends an Event. This method returns the ID of the sent event, or None if the event could not be sent.

pub fn send_event_default<E>(&mut self) -> Option<EventId<E>>

where E: Event + Default,

Sends the default value of the Event of type E. This method returns the ID of the sent event, or None if the event could not be sent.

pub fn send_event_batch<E>( &mut self, events: impl IntoIterator<Item = E>, ) -> Option<SendBatchIds<E>>

where E: Event,

Sends a batch of Events from an iterator. This method returns the IDs of the sent events, or None if the event could not be sent.

pub unsafe fn insert_resource_by_id( &mut self, component_id: ComponentId, value: OwningPtr<'_>, caller: &'static Location<'_>, )

Inserts a new resource with the given value. Will replace the value if it already existed.

You should prefer to use the typed API World::insert_resource where possible and only use this in cases where the actual types are not known at compile time.

Safety

The value referenced by value must be valid for the given ComponentId of this world.

pub unsafe fn insert_non_send_by_id( &mut self, component_id: ComponentId, value: OwningPtr<'_>, caller: &'static Location<'_>, )

Inserts a new !Send resource with the given value. Will replace the value if it already existed.

You should prefer to use the typed API World::insert_non_send_resource where possible and only use this in cases where the actual types are not known at compile time.

Panics

If a value is already present, this function will panic if not called from the same thread that the original value was inserted from.

Safety

The value referenced by value must be valid for the given ComponentId of this world.

pub fn flush(&mut self)

Flushes queued entities and commands.

Queued entities will be spawned, and then commands will be applied.

pub fn increment_change_tick(&mut self) -> Tick

Increments the world’s current change tick and returns the old value.

If you need to call this method, but do not have &mut access to the world, consider using as_unsafe_world_cell_readonly to obtain an UnsafeWorldCell and calling increment_change_tick on that. Note that this can be done in safe code, despite the name of the type.

pub fn read_change_tick(&self) -> Tick

Reads the current change tick of this world.

If you have exclusive (&mut) access to the world, consider using change_tick(), which is more efficient since it does not require atomic synchronization.

pub fn change_tick(&mut self) -> Tick

Reads the current change tick of this world.

This does the same thing as read_change_tick(), only this method is more efficient since it does not require atomic synchronization.

pub fn last_change_tick(&self) -> Tick

When called from within an exclusive system (a System that takes &mut World as its first parameter), this method returns the Tick indicating the last time the exclusive system was run.

Otherwise, this returns the Tick indicating the last time that World::clear_trackers was called.

pub fn last_change_tick_scope<T>( &mut self, last_change_tick: Tick, f: impl FnOnce(&mut World) -> T, ) -> T

Sets World::last_change_tick() to the specified value during a scope. When the scope terminates, it will return to its old value.

This is useful if you need a region of code to be able to react to earlier changes made in the same system.

Examples

// This function runs an update loop repeatedly, allowing each iteration of the loop
// to react to changes made in the previous loop iteration.
fn update_loop(
    world: &mut World,
    mut update_fn: impl FnMut(&mut World) -> std::ops::ControlFlow<()>,
) {
    let mut last_change_tick = world.last_change_tick();

    // Repeatedly run the update function until it requests a break.
    loop {
        let control_flow = world.last_change_tick_scope(last_change_tick, |world| {
            // Increment the change tick so we can detect changes from the previous update.
            last_change_tick = world.change_tick();
            world.increment_change_tick();

            // Update once.
            update_fn(world)
        });

        // End the loop when the closure returns `ControlFlow::Break`.
        if control_flow.is_break() {
            break;
        }
    }
}

pub fn check_change_ticks(&mut self)

Iterates all component change ticks and clamps any older than MAX_CHANGE_AGE. This prevents overflow and thus prevents false positives.

Note: Does nothing if the World counter has not been incremented at least CHECK_TICK_THRESHOLD times since the previous pass.

pub fn clear_all(&mut self)

Runs both clear_entities and clear_resources, invalidating all Entity and resource fetches such as Res, ResMut

pub fn clear_entities(&mut self)

Despawns all entities in this World.

pub fn clear_resources(&mut self)

Clears all resources in this World.

Note: Any resource fetch to this World will fail unless they are re-initialized, including engine-internal resources that are only initialized on app/world construction.

This can easily cause systems expecting certain resources to immediately start panicking. Use with caution.

pub fn register_bundle<B>(&mut self) -> &BundleInfo

where B: Bundle,

Registers all of the components in the given Bundle and returns both the component ids and the bundle id.

This is largely equivalent to calling register_component on each component in the bundle.

impl World

pub fn get_resource_by_id(&self, component_id: ComponentId) -> Option<Ptr<'_>>

Gets a pointer to the resource with the id ComponentId if it exists. The returned pointer must not be used to modify the resource, and must not be dereferenced after the immutable borrow of the World ends.

You should prefer to use the typed API World::get_resource where possible and only use this in cases where the actual types are not known at compile time.

pub fn get_resource_mut_by_id( &mut self, component_id: ComponentId, ) -> Option<MutUntyped<'_>>

Gets a pointer to the resource with the id ComponentId if it exists. The returned pointer may be used to modify the resource, as long as the mutable borrow of the World is still valid.

You should prefer to use the typed API World::get_resource_mut where possible and only use this in cases where the actual types are not known at compile time.

pub fn iter_resources(&self) -> impl Iterator<Item = (&ComponentInfo, Ptr<'_>)>

Iterates over all resources in the world.

The returned iterator provides lifetimed, but type-unsafe pointers. Actually reading the contents of each resource will require the use of unsafe code.

Examples

Printing the size of all resources

let mut total = 0;
for (info, _) in world.iter_resources() {
   println!("Resource: {}", info.name());
   println!("Size: {} bytes", info.layout().size());
   total += info.layout().size();
}
println!("Total size: {} bytes", total);

Dynamically running closures for resources matching specific TypeIds

// In this example, `A` and `B` are resources. We deliberately do not use the
// `bevy_reflect` crate here to showcase the low-level [`Ptr`] usage. You should
// probably use something like `ReflectFromPtr` in a real-world scenario.

// Create the hash map that will store the closures for each resource type
let mut closures: HashMap<TypeId, Box<dyn Fn(&Ptr<'_>)>> = HashMap::new();

// Add closure for `A`
closures.insert(TypeId::of::<A>(), Box::new(|ptr| {
    // SAFETY: We assert ptr is the same type of A with TypeId of A
    let a = unsafe { &ptr.deref::<A>() };
    // ... do something with `a` here
}));

// Add closure for `B`
closures.insert(TypeId::of::<B>(), Box::new(|ptr| {
    // SAFETY: We assert ptr is the same type of B with TypeId of B
    let b = unsafe { &ptr.deref::<B>() };
    // ... do something with `b` here
}));

// Iterate all resources, in order to run the closures for each matching resource type
for (info, ptr) in world.iter_resources() {
    let Some(type_id) = info.type_id() else {
       // It's possible for resources to not have a `TypeId` (e.g. non-Rust resources
       // dynamically inserted via a scripting language) in which case we can't match them.
       continue;
    };

    let Some(closure) = closures.get(&type_id) else {
       // No closure for this resource type, skip it.
       continue;
    };

    // Run the closure for the resource
    closure(&ptr);
}

pub fn iter_resources_mut( &mut self, ) -> impl Iterator<Item = (&ComponentInfo, MutUntyped<'_>)>

Mutably iterates over all resources in the world.

The returned iterator provides lifetimed, but type-unsafe pointers. Actually reading from or writing to the contents of each resource will require the use of unsafe code.

Example

// In this example, `A` and `B` are resources. We deliberately do not use the
// `bevy_reflect` crate here to showcase the low-level `MutUntyped` usage. You should
// probably use something like `ReflectFromPtr` in a real-world scenario.

// Create the hash map that will store the mutator closures for each resource type
let mut mutators: HashMap<TypeId, Box<dyn Fn(&mut MutUntyped<'_>)>> = HashMap::new();

// Add mutator closure for `A`
mutators.insert(TypeId::of::<A>(), Box::new(|mut_untyped| {
    // Note: `MutUntyped::as_mut()` automatically marks the resource as changed
    // for ECS change detection, and gives us a `PtrMut` we can use to mutate the resource.
    // SAFETY: We assert ptr is the same type of A with TypeId of A
    let a = unsafe { &mut mut_untyped.as_mut().deref_mut::<A>() };
    // ... mutate `a` here
}));

// Add mutator closure for `B`
mutators.insert(TypeId::of::<B>(), Box::new(|mut_untyped| {
    // SAFETY: We assert ptr is the same type of B with TypeId of B
    let b = unsafe { &mut mut_untyped.as_mut().deref_mut::<B>() };
    // ... mutate `b` here
}));

// Iterate all resources, in order to run the mutator closures for each matching resource type
for (info, mut mut_untyped) in world.iter_resources_mut() {
    let Some(type_id) = info.type_id() else {
       // It's possible for resources to not have a `TypeId` (e.g. non-Rust resources
       // dynamically inserted via a scripting language) in which case we can't match them.
       continue;
    };

    let Some(mutator) = mutators.get(&type_id) else {
       // No mutator closure for this resource type, skip it.
       continue;
    };

    // Run the mutator closure for the resource
    mutator(&mut mut_untyped);
}

pub fn get_non_send_by_id(&self, component_id: ComponentId) -> Option<Ptr<'_>>

Gets a !Send resource to the resource with the id ComponentId if it exists. The returned pointer must not be used to modify the resource, and must not be dereferenced after the immutable borrow of the World ends.

You should prefer to use the typed API World::get_resource where possible and only use this in cases where the actual types are not known at compile time.

Panics

This function will panic if it isn’t called from the same thread that the resource was inserted from.

pub fn get_non_send_mut_by_id( &mut self, component_id: ComponentId, ) -> Option<MutUntyped<'_>>

Gets a !Send resource to the resource with the id ComponentId if it exists. The returned pointer may be used to modify the resource, as long as the mutable borrow of the World is still valid.

You should prefer to use the typed API World::get_resource_mut where possible and only use this in cases where the actual types are not known at compile time.

Panics

This function will panic if it isn’t called from the same thread that the resource was inserted from.

pub fn remove_resource_by_id(&mut self, component_id: ComponentId) -> Option<()>

Removes the resource of a given type, if it exists. Otherwise returns None.

You should prefer to use the typed API World::remove_resource where possible and only use this in cases where the actual types are not known at compile time.

pub fn remove_non_send_by_id(&mut self, component_id: ComponentId) -> Option<()>

Removes the resource of a given type, if it exists. Otherwise returns None.

You should prefer to use the typed API World::remove_resource where possible and only use this in cases where the actual types are not known at compile time.

Panics

This function will panic if it isn’t called from the same thread that the resource was inserted from.

pub fn get_by_id( &self, entity: Entity, component_id: ComponentId, ) -> Option<Ptr<'_>>

Retrieves an immutable untyped reference to the given entity’s Component of the given ComponentId. Returns None if the entity does not have a Component of the given type.

You should prefer to use the typed API World::get_mut where possible and only use this in cases where the actual types are not known at compile time.

Panics

This function will panic if it isn’t called from the same thread that the resource was inserted from.

pub fn get_mut_by_id( &mut self, entity: Entity, component_id: ComponentId, ) -> Option<MutUntyped<'_>>

Retrieves a mutable untyped reference to the given entity’s Component of the given ComponentId. Returns None if the entity does not have a Component of the given type.

You should prefer to use the typed API World::get_mut where possible and only use this in cases where the actual types are not known at compile time.

impl World

pub fn add_schedule(&mut self, schedule: Schedule)

Adds the specified Schedule to the world. The schedule can later be run by calling .run_schedule(label) or by directly accessing the Schedules resource.

The Schedules resource will be initialized if it does not already exist.

pub fn try_schedule_scope<R>( &mut self, label: impl ScheduleLabel, f: impl FnOnce(&mut World, &mut Schedule) -> R, ) -> Result<R, TryRunScheduleError>

Temporarily removes the schedule associated with label from the world, runs user code, and finally re-adds the schedule. This returns a TryRunScheduleError if there is no schedule associated with label.

The Schedule is fetched from the Schedules resource of the world by its label, and system state is cached.

For simple cases where you just need to call the schedule once, consider using World::try_run_schedule instead. For other use cases, see the example on World::schedule_scope.

pub fn schedule_scope<R>( &mut self, label: impl ScheduleLabel, f: impl FnOnce(&mut World, &mut Schedule) -> R, ) -> R

Temporarily removes the schedule associated with label from the world, runs user code, and finally re-adds the schedule.

The Schedule is fetched from the Schedules resource of the world by its label, and system state is cached.

Examples

// Run the schedule five times.
world.schedule_scope(MySchedule, |world, schedule| {
    for _ in 0..5 {
        schedule.run(world);
    }
});

For simple cases where you just need to call the schedule once, consider using World::run_schedule instead.

Panics

If the requested schedule does not exist.

pub fn try_run_schedule( &mut self, label: impl ScheduleLabel, ) -> Result<(), TryRunScheduleError>

Attempts to run the Schedule associated with the label a single time, and returns a TryRunScheduleError if the schedule does not exist.

The Schedule is fetched from the Schedules resource of the world by its label, and system state is cached.

For simple testing use cases, call Schedule::run(&mut world) instead.

Examples found in repository

examples/state/custom_transitions.rs (line 113)

    fn run_reenter<S: States>(transition: In<Option<StateTransitionEvent<S>>>, world: &mut World) {
        // We return early if no transition event happened.
        let Some(transition) = transition.0 else {
            return;
        };

        // If we wanted to ignore identity transitions,
        // we'd compare `exited` and `entered` here,
        // and return if they were the same.

        // We check if we actually entered a state.
        // A [`None`] would indicate that the state was removed from the world.
        // This only happens in the case of [`SubStates`] and [`ComputedStates`].
        let Some(entered) = transition.entered else {
            return;
        };

        // If all conditions are valid, we run our custom schedule.
        let _ = world.try_run_schedule(OnReenter(entered));

        // If you want to overwrite the default `OnEnter` behavior to act like re-enter,
        // you can do so by running the `OnEnter` schedule here. Note that you don't want
        // to run `OnEnter` when the default behavior does so.
        // ```
        // if transition.entered != transition.exited {
        //     return;
        // }
        // let _ = world.try_run_schedule(OnReenter(entered));
        // ```
    }

    /// Custom schedule that will behave like [`OnExit`], but run on identity transitions.
    #[derive(ScheduleLabel, Clone, Debug, PartialEq, Eq, Hash)]
    pub struct OnReexit<S: States>(pub S);

    fn run_reexit<S: States>(transition: In<Option<StateTransitionEvent<S>>>, world: &mut World) {
        let Some(transition) = transition.0 else {
            return;
        };
        let Some(exited) = transition.exited else {
            return;
        };

        let _ = world.try_run_schedule(OnReexit(exited));
    }

pub fn run_schedule(&mut self, label: impl ScheduleLabel)

Runs the Schedule associated with the label a single time.

The Schedule is fetched from the Schedules resource of the world by its label, and system state is cached.

For simple testing use cases, call Schedule::run(&mut world) instead.

Panics

If the requested schedule does not exist.

pub fn allow_ambiguous_component<T>(&mut self)

where T: Component,

Ignore system order ambiguities caused by conflicts on Components of type T.

pub fn allow_ambiguous_resource<T>(&mut self)

where T: Resource,

Ignore system order ambiguities caused by conflicts on Resources of type T.

Trait Implementations

impl Debug for World

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl Default for World

fn default() -> World

Returns the “default value” for a type.

impl DirectAssetAccessExt for World

fn add_asset<'a, A>(&mut self, asset: impl Into<A>) -> Handle<A>

where A: Asset,

Insert an asset similarly to Assets::add.

Panics

If self doesn’t have an AssetServer resource initialized yet.

fn load_asset<'a, A>(&self, path: impl Into<AssetPath<'a>>) -> Handle<A>

where A: Asset,

Load an asset similarly to AssetServer::load.

Panics

If self doesn’t have an AssetServer resource initialized yet.

fn load_asset_with_settings<'a, A, S>( &self, path: impl Into<AssetPath<'a>>, settings: impl Fn(&mut S) + Send + Sync + 'static, ) -> Handle<A>

where A: Asset, S: Settings,

Load an asset with settings, similarly to AssetServer::load_with_settings.

Panics

If self doesn’t have an AssetServer resource initialized yet.

impl Drop for World

fn drop(&mut self)

Executes the destructor for this type.

impl<'w> From<&'w World> for FilteredResources<'w, 'static>

fn from(value: &'w World) -> FilteredResources<'w, 'static>

Converts to this type from the input type.

impl<'w> From<&'w World> for UnsafeWorldCell<'w>

fn from(value: &'w World) -> UnsafeWorldCell<'w>

Converts to this type from the input type.

impl<'w> From<&'w mut World> for DeferredWorld<'w>

fn from(world: &'w mut World) -> DeferredWorld<'w>

Converts to this type from the input type.

impl<'w> From<&'w mut World> for FilteredResources<'w, 'static>

fn from(value: &'w mut World) -> FilteredResources<'w, 'static>

Converts to this type from the input type.

impl<'w> From<&'w mut World> for FilteredResourcesMut<'w, 'static>

fn from(value: &'w mut World) -> FilteredResourcesMut<'w, 'static>

Converts to this type from the input type.

impl<'w> From<&'w mut World> for UnsafeWorldCell<'w>

fn from(value: &'w mut World) -> UnsafeWorldCell<'w>

Converts to this type from the input type.

impl RunSystemOnce for &mut World

fn run_system_once_with<T, In, Out, Marker>( self, input: <<<T as IntoSystem<In, Out, Marker>>::System as System>::In as SystemInput>::Inner<'_>, system: T, ) -> Result<Out, RunSystemError>

where T: IntoSystem<In, Out, Marker>, In: SystemInput,

Tries to run a system with given input and apply deferred parameters.

fn run_system_once<T, Out, Marker>( self, system: T, ) -> Result<Out, RunSystemError>

where T: IntoSystem<(), Out, Marker>,

Tries to run a system and apply its deferred parameters.

impl SystemParam for &World

type State = ()

Used to store data which persists across invocations of a system.

type Item<'w, 's> = &'w World

The item type returned when constructing this system param. The value of this associated type should be Self, instantiated with new lifetimes.

fn init_state( _world: &mut World, system_meta: &mut SystemMeta, ) -> <&World as SystemParam>::State

Registers any World access used by this SystemParam and creates a new instance of this param’s State.

unsafe fn get_param<'w, 's>( _state: &'s mut <&World as SystemParam>::State, _system_meta: &SystemMeta, world: UnsafeWorldCell<'w>, _change_tick: Tick, ) -> <&World as SystemParam>::Item<'w, 's>

Creates a parameter to be passed into a SystemParamFunction.

unsafe fn new_archetype( state: &mut Self::State, archetype: &Archetype, system_meta: &mut SystemMeta, )

For the specified Archetype, registers the components accessed by this SystemParam (if applicable).a

fn apply(state: &mut Self::State, system_meta: &SystemMeta, world: &mut World)

Applies any deferred mutations stored in this SystemParam’s state. This is used to apply Commands during apply_deferred.

fn queue( state: &mut Self::State, system_meta: &SystemMeta, world: DeferredWorld<'_>, )

Queues any deferred mutations to be applied at the next apply_deferred.

unsafe fn validate_param( _state: &Self::State, _system_meta: &SystemMeta, _world: UnsafeWorldCell<'_>, ) -> bool

Validates that the param can be acquired by the get_param. Built-in executors use this to prevent systems with invalid params from running. For nested SystemParams validation will fail if any delegated validation fails.

impl<'w> ReadOnlySystemParam for &'w World

SAFETY: only reads world

impl Send for World

impl Sync for World