Trait Component
pub trait Component:
Send
+ Sync
+ 'static {
type Mutability: ComponentMutability;
const STORAGE_TYPE: StorageType;
// Provided methods
fn on_add() -> Option<for<'w> fn(DeferredWorld<'w>, HookContext)> { ... }
fn on_insert() -> Option<for<'w> fn(DeferredWorld<'w>, HookContext)> { ... }
fn on_discard() -> Option<for<'w> fn(DeferredWorld<'w>, HookContext)> { ... }
fn on_remove() -> Option<for<'w> fn(DeferredWorld<'w>, HookContext)> { ... }
fn on_despawn() -> Option<for<'w> fn(DeferredWorld<'w>, HookContext)> { ... }
fn register_required_components(
_component_id: ComponentId,
_required_components: &mut RequiredComponentsRegistrator<'_, '_>,
) { ... }
fn clone_behavior() -> ComponentCloneBehavior { ... }
fn map_entities<E>(_this: &mut Self, _mapper: &mut E)
where E: EntityMapper { ... }
fn relationship_accessor() -> Option<ComponentRelationshipAccessor<Self>> { ... }
}Expand description
A data type that can be used to store data for an entity.
Component is a derivable trait: this means that a data type can implement it by applying a #[derive(Component)] attribute to it.
However, components must always satisfy the Send + Sync + 'static trait bounds.
§Examples
Components can take many forms: they are usually structs, but can also be of every other kind of data type, like enums or zero sized types. The following examples show how components are laid out in code.
// A component can contain data...
#[derive(Component)]
struct LicensePlate(String);
// ... but it can also be a zero-sized marker.
#[derive(Component)]
struct Car;
// Components can also be structs with named fields...
#[derive(Component)]
struct VehiclePerformance {
acceleration: f32,
top_speed: f32,
handling: f32,
}
// ... or enums.
#[derive(Component)]
enum WheelCount {
Two,
Three,
Four,
}§Component and data access
Components can be marked as immutable by adding the #[component(immutable)]
attribute when using the derive macro.
See the documentation for ComponentMutability for more details around this
feature.
See the entity module level documentation to learn how to add or remove components from an entity.
See the documentation for Query to learn how to access component data from a system.
§Choosing a storage type
Components can be stored in the world using different strategies with their own performance implications.
By default, components are added to the Table storage, which is optimized for query iteration.
Alternatively, components can be added to the SparseSet storage, which is optimized for component insertion and removal.
This is achieved by adding an additional #[component(storage = "SparseSet")] attribute to the derive one:
#[derive(Component)]
#[component(storage = "SparseSet")]
struct ComponentA;§Required Components
Components can specify Required Components. If some Component A requires Component B, then when A is inserted,
B will also be initialized and inserted (if it was not manually specified).
The Default constructor will be used to initialize the component, by default:
#[derive(Component)]
#[require(B)]
struct A;
#[derive(Component, Default, PartialEq, Eq, Debug)]
struct B(usize);
// This will implicitly also insert B with the Default constructor
let id = world.spawn(A).id();
assert_eq!(&B(0), world.entity(id).get::<B>().unwrap());
// This will _not_ implicitly insert B, because it was already provided
world.spawn((A, B(11)));Components can have more than one required component:
#[derive(Component)]
#[require(B, C)]
struct A;
#[derive(Component, Default, PartialEq, Eq, Debug)]
#[require(C)]
struct B(usize);
#[derive(Component, Default, PartialEq, Eq, Debug)]
struct C(u32);
// 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());You can define inline component values that take the following forms:
#[derive(Component)]
#[require(
B(1), // tuple structs
C { // named-field structs
x: 1,
..Default::default()
},
D::One, // enum variants
E::ONE, // associated consts
F::new(1) // constructors
)]
struct A;
#[derive(Component, PartialEq, Eq, Debug)]
struct B(u8);
#[derive(Component, PartialEq, Eq, Debug, Default)]
struct C {
x: u8,
y: u8,
}
#[derive(Component, PartialEq, Eq, Debug)]
enum D {
Zero,
One,
}
#[derive(Component, PartialEq, Eq, Debug)]
struct E(u8);
impl E {
pub const ONE: Self = Self(1);
}
#[derive(Component, PartialEq, Eq, Debug)]
struct F(u8);
impl F {
fn new(value: u8) -> Self {
Self(value)
}
}
let id = world.spawn(A).id();
assert_eq!(&B(1), world.entity(id).get::<B>().unwrap());
assert_eq!(&C { x: 1, y: 0 }, world.entity(id).get::<C>().unwrap());
assert_eq!(&D::One, world.entity(id).get::<D>().unwrap());
assert_eq!(&E(1), world.entity(id).get::<E>().unwrap());
assert_eq!(&F(1), world.entity(id).get::<F>().unwrap());You can also define arbitrary expressions by using =
#[derive(Component)]
#[require(C = init_c())]
struct A;
#[derive(Component, PartialEq, Eq, Debug)]
#[require(C = C(20))]
struct B;
#[derive(Component, PartialEq, Eq, Debug)]
struct C(usize);
fn init_c() -> C {
C(10)
}
// This will implicitly also insert C with the init_c() constructor
let id = world.spawn(A).id();
assert_eq!(&C(10), world.entity(id).get::<C>().unwrap());
// This will implicitly also insert C with the `|| C(20)` constructor closure
let id = world.spawn(B).id();
assert_eq!(&C(20), world.entity(id).get::<C>().unwrap());Required components are recursive. This means, if a Required Component has required components, those components will also be inserted if they are missing:
#[derive(Component)]
#[require(B)]
struct A;
#[derive(Component, Default, PartialEq, Eq, Debug)]
#[require(C)]
struct B(usize);
#[derive(Component, Default, PartialEq, Eq, Debug)]
struct C(u32);
// 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());Note that cycles in the “component require tree” will result in stack overflows when attempting to insert a component.
This “multiple inheritance” pattern does mean that it is possible to have duplicate requires for a given type
at different levels of the inheritance tree:
#[derive(Component)]
struct X(usize);
#[derive(Component, Default)]
#[require(X(1))]
struct Y;
#[derive(Component)]
#[require(
Y,
X(2),
)]
struct Z;
// In this case, the x2 constructor is used for X
let id = world.spawn(Z).id();
assert_eq!(2, world.entity(id).get::<X>().unwrap().0);In general, this shouldn’t happen often, but when it does the algorithm for choosing the constructor from the tree is simple and predictable:
- A constructor from a direct
#[require()], if one exists, is selected with priority. - Otherwise, perform a Depth First Search on the tree of requirements and select the first one found.
From a user perspective, just think about this as the following:
- Specifying a required component constructor for Foo directly on a spawned component Bar will result in that constructor being used (and overriding existing constructors lower in the inheritance tree). This is the classic “inheritance override” behavior people expect.
- For cases where “multiple inheritance” results in constructor clashes, Components should be listed in “importance order”. List a component earlier in the requirement list to initialize its inheritance tree earlier.
§Registering required components at runtime
In most cases, required components should be registered using the require attribute as shown above.
However, in some cases, it may be useful to register required components at runtime.
This can be done through World::register_required_components or World::register_required_components_with
for the Default and custom constructors respectively:
#[derive(Component)]
struct A;
#[derive(Component, Default, PartialEq, Eq, Debug)]
struct B(usize);
#[derive(Component, 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_with::<B, C>(|| C(2));
// This will implicitly also insert B with its Default constructor
// and C with the custom constructor defined by B.
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());Similar rules as before apply to duplicate requires for a given type at different levels
of the inheritance tree. A requiring C directly would take precedence over indirectly
requiring it through A requiring B and B requiring C.
Unlike with the require attribute, directly requiring the same component multiple times
for the same component will result in a panic. This is done to prevent conflicting constructors
and confusing ordering dependencies.
Note that requirements must currently be registered before the requiring component is inserted into the world for the first time. Registering requirements after this will lead to a panic.
§Relationships between Entities
Sometimes it is useful to define relationships between entities. A common example is the
parent / child relationship. Since Components are how data is stored for Entities, one might
naturally think to create a Component which has a field of type Entity.
To facilitate this pattern, Bevy provides the Relationship
trait. You can derive the Relationship and
RelationshipTarget traits in addition to the
Component trait in order to implement data driven relationships between entities, see the trait
docs for more details.
In addition, Bevy provides canonical implementations of the parent / child relationship via the
ChildOf Relationship and
the Children
RelationshipTarget.
§Adding component’s hooks
See ComponentHooks for a detailed explanation of component’s hooks.
Alternatively to the example shown in ComponentHooks’ documentation, hooks can be configured using following attributes:
#[component(on_add = on_add_function)]#[component(on_insert = on_insert_function)]#[component(on_discard = on_discard_function)]#[component(on_remove = on_remove_function)]
#[derive(Component)]
#[component(on_add = my_on_add_hook)]
#[component(on_insert = my_on_insert_hook)]
// Another possible way of configuring hooks:
// #[component(on_add = my_on_add_hook, on_insert = my_on_insert_hook)]
//
// We don't have a discard or remove hook, so we can leave them out:
// #[component(on_discard = my_on_discard_hook, on_remove = my_on_remove_hook)]
struct ComponentA;
fn my_on_add_hook(world: DeferredWorld, context: HookContext) {
// ...
}
// You can also destructure items directly in the signature
fn my_on_insert_hook(world: DeferredWorld, HookContext { caller, .. }: HookContext) {
// ...
}This also supports function calls that yield closures
#[derive(Component)]
#[component(on_add = my_msg_hook("hello"))]
#[component(on_despawn = my_msg_hook("yoink"))]
struct ComponentA;
// a hook closure generating function
fn my_msg_hook(message: &'static str) -> impl Fn(DeferredWorld, HookContext) {
move |_world, _ctx| {
println!("{message}");
}
}
A hook’s function path can be elided if it is Self::on_add, Self::on_insert etc.
#[derive(Component, Debug)]
#[component(on_add)]
struct DoubleOnSpawn(usize);
impl DoubleOnSpawn {
fn on_add(mut world: DeferredWorld, context: HookContext) {
let mut entity = world.get_mut::<Self>(context.entity).unwrap();
entity.0 *= 2;
}
}§Setting the clone behavior
You can specify how the Component is cloned when deriving it.
Your options are the functions and variants of [ComponentCloneBehavior]
See Clone Behaviors section of EntityCloner to understand how this affects handler priority.
#[derive(Component)]
#[component(clone_behavior = Ignore)]
struct MyComponent;
§Implementing the trait for foreign types
As a consequence of the orphan rule, it is not possible to separate into two different crates the implementation of Component from the definition of a type.
This means that it is not possible to directly have a type defined in a third party library as a component.
This important limitation can be easily worked around using the newtype pattern:
this makes it possible to locally define and implement Component for a tuple struct that wraps the foreign type.
The following example gives a demonstration of this pattern.
// `Component` is defined in the `bevy_ecs` crate.
use bevy_ecs::component::Component;
// `Duration` is defined in the `std` crate.
use std::time::Duration;
// It is not possible to implement `Component` for `Duration` from this position, as they are
// both foreign items, defined in an external crate. However, nothing prevents to define a new
// `Cooldown` type that wraps `Duration`. As `Cooldown` is defined in a local crate, it is
// possible to implement `Component` for it.
#[derive(Component)]
struct Cooldown(Duration);§!Sync Components
A !Sync type cannot implement Component. However, it is possible to wrap a Send but not Sync
type in SyncCell or the currently unstable Exclusive to make it Sync. This forces only
having mutable access (&mut T only, never &T), but makes it safe to reference across multiple
threads.
This will fail to compile since RefCell is !Sync.
#[derive(Component)]
struct NotSync {
counter: RefCell<usize>,
}This will compile since the RefCell is wrapped with SyncCell.
use bevy_platform::cell::SyncCell;
// This will compile.
#[derive(Component)]
struct ActuallySync {
counter: SyncCell<RefCell<usize>>,
}Required Associated Constants§
const STORAGE_TYPE: StorageType
const STORAGE_TYPE: StorageType
A constant indicating the storage type used for this component.
Required Associated Types§
type Mutability: ComponentMutability
type Mutability: ComponentMutability
A marker type to assist Bevy with determining if this component is
mutable, or immutable. Mutable components will have Component<Mutability = Mutable>,
while immutable components will instead have Component<Mutability = Immutable>.
- For a component to be mutable, this type must be [
Mutable]. - For a component to be immutable, this type must be [
Immutable].
Provided Methods§
fn on_add() -> Option<for<'w> fn(DeferredWorld<'w>, HookContext)>
fn on_add() -> Option<for<'w> fn(DeferredWorld<'w>, HookContext)>
Gets the on_add [ComponentHook] for this Component if one is defined.
fn on_insert() -> Option<for<'w> fn(DeferredWorld<'w>, HookContext)>
fn on_insert() -> Option<for<'w> fn(DeferredWorld<'w>, HookContext)>
Gets the on_insert [ComponentHook] for this Component if one is defined.
fn on_discard() -> Option<for<'w> fn(DeferredWorld<'w>, HookContext)>
fn on_discard() -> Option<for<'w> fn(DeferredWorld<'w>, HookContext)>
Gets the on_discard [ComponentHook] for this Component if one is defined.
fn on_remove() -> Option<for<'w> fn(DeferredWorld<'w>, HookContext)>
fn on_remove() -> Option<for<'w> fn(DeferredWorld<'w>, HookContext)>
Gets the on_remove [ComponentHook] for this Component if one is defined.
fn on_despawn() -> Option<for<'w> fn(DeferredWorld<'w>, HookContext)>
fn on_despawn() -> Option<for<'w> fn(DeferredWorld<'w>, HookContext)>
Gets the on_despawn [ComponentHook] for this Component if one is defined.
fn register_required_components(
_component_id: ComponentId,
_required_components: &mut RequiredComponentsRegistrator<'_, '_>,
)
fn register_required_components( _component_id: ComponentId, _required_components: &mut RequiredComponentsRegistrator<'_, '_>, )
Registers required components.
§Safety
_required_componentsmust only contain components valid in_components.
fn clone_behavior() -> ComponentCloneBehavior
fn clone_behavior() -> ComponentCloneBehavior
Called when registering this component, allowing to override clone function (or disable cloning altogether) for this component.
See Clone Behaviors section of EntityCloner to understand how this affects handler priority.
fn map_entities<E>(_this: &mut Self, _mapper: &mut E)where
E: EntityMapper,
fn map_entities<E>(_this: &mut Self, _mapper: &mut E)where
E: EntityMapper,
Maps the entities on this component using the given EntityMapper. This is used to remap entities in contexts like scenes and entity cloning.
When deriving Component, this is populated by annotating fields containing entities with #[entities]
#[derive(Component)]
struct Inventory {
#[entities]
items: Vec<Entity>
}Fields with #[entities] must implement MapEntities.
Bevy provides various implementations of MapEntities, so that arbitrary combinations like these are supported with #[entities]:
#[derive(Component)]
struct Inventory {
#[entities]
items: Vec<Option<Entity>>
}You might need more specialized logic. A likely cause of this is your component contains collections of entities that
don’t implement MapEntities. In that case, you can annotate your component with
#[component(map_entities)]. Using this attribute, you must implement MapEntities for the
component itself, and this method will simply call that implementation.
#[derive(Component)]
#[component(map_entities)]
struct Inventory {
items: EntityHashMap<usize>
}
impl MapEntities for Inventory {
fn map_entities<M: EntityMapper>(&mut self, entity_mapper: &mut M) {
self.items = self.items
.drain()
.map(|(id, count)|(entity_mapper.get_mapped(id), count))
.collect();
}
}Alternatively, you can specify the path to a function with #[component(map_entities = function_path)], similar to component hooks.
In this case, the inputs of the function should mirror the inputs to this method, with the second parameter being generic.
#[derive(Component)]
#[component(map_entities = map_the_map)]
// Also works: map_the_map::<M> or map_the_map::<_>
struct Inventory {
items: EntityHashMap<usize>
}
fn map_the_map<M: EntityMapper>(inv: &mut Inventory, entity_mapper: &mut M) {
inv.items = inv.items
.drain()
.map(|(id, count)|(entity_mapper.get_mapped(id), count))
.collect();
}You can use the turbofish (::<A,B,C>) to specify parameters when a function is generic, using either M or _ for the type of the mapper parameter.
fn relationship_accessor() -> Option<ComponentRelationshipAccessor<Self>>
fn relationship_accessor() -> Option<ComponentRelationshipAccessor<Self>>
Returns [ComponentRelationshipAccessor] required for working with relationships in dynamic contexts.
If component is not a Relationship or RelationshipTarget, this should return None.
Dyn Compatibility§
This trait is not dyn compatible.
In older versions of Rust, dyn compatibility was called "object safety", so this trait is not object safe.