# Objective
- Provide an expressive way to register dynamic behavior in response to
ECS changes that is consistent with existing bevy types and traits as to
provide a smooth user experience.
- Provide a mechanism for immediate changes in response to events during
command application in order to facilitate improved query caching on the
path to relations.
## Solution
- A new fundamental ECS construct, the `Observer`; inspired by flec's
observers but adapted to better fit bevy's access patterns and rust's
type system.
---
## Examples
There are 3 main ways to register observers. The first is a "component
observer" that looks like this:
```rust
world.observe(|trigger: Trigger<OnAdd, Transform>, query: Query<&Transform>| {
let transform = query.get(trigger.entity()).unwrap();
});
```
The above code will spawn a new entity representing the observer that
will run it's callback whenever the `Transform` component is added to an
entity. This is a system-like function that supports dependency
injection for all the standard bevy types: `Query`, `Res`, `Commands`
etc. It also has a `Trigger` parameter that provides information about
the trigger such as the target entity, and the event being triggered.
Importantly these systems run during command application which is key
for their future use to keep ECS internals up to date. There are similar
events for `OnInsert` and `OnRemove`, and this will be expanded with
things such as `ArchetypeCreated`, `TableEmpty` etc. in follow up PRs.
Another way to register an observer is an "entity observer" that looks
like this:
```rust
world.entity_mut(entity).observe(|trigger: Trigger<Resize>| {
// ...
});
```
Entity observers run whenever an event of their type is triggered
targeting that specific entity. This type of observer will de-spawn
itself if the entity (or entities) it is observing is ever de-spawned so
as to not leave dangling observers.
Entity observers can also be spawned from deferred contexts such as
other observers, systems, or hooks using commands:
```rust
commands.entity(entity).observe(|trigger: Trigger<Resize>| {
// ...
});
```
Observers are not limited to in built event types, they can be used with
any type that implements `Event` (which has been extended to implement
Component). This means events can also carry data:
```rust
#[derive(Event)]
struct Resize { x: u32, y: u32 }
commands.entity(entity).observe(|trigger: Trigger<Resize>, query: Query<&mut Size>| {
let event = trigger.event();
// ...
});
// Will trigger the observer when commands are applied.
commands.trigger_targets(Resize { x: 10, y: 10 }, entity);
```
You can also trigger events that target more than one entity at a time:
```rust
commands.trigger_targets(Resize { x: 10, y: 10 }, [e1, e2]);
```
Additionally, Observers don't _need_ entity targets:
```rust
app.observe(|trigger: Trigger<Quit>| {
})
commands.trigger(Quit);
```
In these cases, `trigger.entity()` will be a placeholder.
Observers are actually just normal entities with an `ObserverState` and
`Observer` component! The `observe()` functions above are just shorthand
for:
```rust
world.spawn(Observer::new(|trigger: Trigger<Resize>| {});
```
This will spawn the `Observer` system and use an `on_add` hook to add
the `ObserverState` component.
Dynamic components and trigger types are also fully supported allowing
for runtime defined trigger types.
## Possible Follow-ups
1. Deprecate `RemovedComponents`, observers should fulfill all use cases
while being more flexible and performant.
2. Queries as entities: Swap queries to entities and begin using
observers listening to archetype creation triggers to keep their caches
in sync, this allows unification of `ObserverState` and `QueryState` as
well as unlocking several API improvements for `Query` and the
management of `QueryState`.
3. Trigger bubbling: For some UI use cases in particular users are
likely to want some form of bubbling for entity observers, this is
trivial to implement naively but ideally this includes an acceleration
structure to cache hierarchy traversals.
4. All kinds of other in-built trigger types.
5. Optimization; in order to not bloat the complexity of the PR I have
kept the implementation straightforward, there are several areas where
performance can be improved. The focus for this PR is to get the
behavior implemented and not incur a performance cost for users who
don't use observers.
I am leaving each of these to follow up PR's in order to keep each of
them reviewable as this already includes significant changes.
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
Co-authored-by: MiniaczQ <xnetroidpl@gmail.com>
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
…izer (#13442)"
This reverts commit 5cfb063d4a.
- This PR broke bevy-trait-query, which needs to be able to write a
resource in init_state. See #13798 for more details.
- Note this doesn't fix everything as transmutes for bevy-trait-query
will still be broken,. But the current usage in that crate is UB, so we
need to find another solution.
# Objective
In #13343, `WorldQuery::get_state` was constrained from `&World` as the
argument to `&Components`, but `WorldQuery::init_state` hasn't yet been
changed from `&mut World` to match.
Fixes#13358
## Solution
Create a wrapper around `&mut Components` and `&mut Storages` that can
be obtained from `&mut World` with a `component_initializer` method.
This new `ComponentInitializer` re-exposes the API on `&mut Components`
minus the `&mut Storages` parameter where it was present. For the
`&Components` API, it simply derefs to its `components` field.
## Changelog
### Added
The `World::component_initializer` method.
The `ComponentInitializer` struct that re-exposes `Components` API.
### Changed
`WorldQuery::init_state` now takes `&mut ComponentInitializer` instead
of `&mut World`.
## Migration Guide
Instead of passing `&mut World` to `WorldQuery::init_state` directly,
pass in a mutable reference to the struct returned from
`World::component_initializer`.
# Objective
The current query iterators cannot be used in positions with a `Debug`
bound.
F.e. when they are packaged in `Result` in the error position, `expect`
cannot be called on them.
Required for `QueryManyIter::entities_all_unique` in #13477.
## Solution
Add simple `Debug` impls that print the query iterator names.
## Changelog
`QueryIter`, `QueryManyIter`, `QueryCombinationIter`, and
`QuerySortedIter` now implement `Debug`.
# Objective
Currently, a query iterator can be collected into a `Vec` and sorted,
but this can be quite unwieldy, especially when many `Component`s are
involved. The `itertools` crate helps somewhat, but the need to write a
closure over all of `QueryData`
can sometimes hurt ergonomics, anywhere from slightly to strongly. A key
extraction function only partially helps, as `sort_by_key` does not
allow returning non-`Copy` data. `sort_by` does not suffer from the
`Copy` restriction, but now the user has to write out a `cmp` function
over two `QueryData::Item`s when it could have just been handled by the
`Ord` impl for the key.
`sort` requires the entire `Iterator` Item to be `Ord`, which is rarely
usable without manual helper functionality. If the user wants to hide
away unused components with a `..` range, they need to track item tuple
order across their function. Mutable `QueryData` can also introduce
further complexity.
Additionally, sometimes users solely include `Component`s /`Entity` to
guarantee iteration order.
For a user to write a function to abstract away repeated sorts over
various `QueryData` types they use would require reaching for the
`all_tuples!` macro, and continue tracking tuple order afterwards.
Fixes https://github.com/bevyengine/bevy/issues/1470.
## Solution
Custom sort methods on `QueryIter`, which take a query lens as a generic
argument, like `transmute_lens` in `Query`.
This allows users to choose what part of their queries they pass to
their sort function calls, serving as a kind of "key extraction
function" before the sort call. F.e. allowing users to implement `Ord`
for a Component, then call `query.iter().sort::<OrdComponent>()`
This works independent of mutability in `QueryData`, `QueryData` tuple
order, or the underlying `iter/iter_mut` call.
Non-`Copy` components could also be used this way, an internal
`Arc<usize>` being an example.
If `Ord` impls on components do not suffice, other sort methods can be
used. Notably useful when combined with `EntityRef` or `EntityMut`.
Another boon from using underlying `transmute` functionality, is that
with the [allowed
transmutes](http://dev-docs.bevyengine.org/bevy/ecs/prelude/struct.Query.html#allowed-transmutes),
it is possible to sort a `Query` with `Entity` even if it wasn't
included in the original `Query`.
The additional generic parameter on the methods other than `sort` and
`sort_unstable` currently cannot be removed due to Rust limitations,
however their types can be inferred.
The new methods do not conflict with the `itertools` sort methods, as
those use the "sorted" prefix.
This is implemented barely touching existing code. That change to
existing code being that `QueryIter` now holds on to the reference to
`UnsafeWorldCell` that is used to initialize it.
A lens query is constructed with `Entity` attached at the end, sorted,
and turned into an iterator. The iterator maps away the lens query,
leaving only an iterator of `Entity`, which is used by `QuerySortedIter`
to retrieve the actual items.
`QuerySortedIter` resembles a combination of `QueryManyIter` and
`QueryIter`, but it uses an entity list that is guaranteed to contain
unique entities, and implements `ExactSizeIterator`,
`DoubleEndedIterator`, `FusedIterator` regardless of mutability or
filter kind (archetypal/non-archetypal).
The sort methods are not allowed to be called after `next`, and will
panic otherwise. This is checked using `QueryIterationCursor` state,
which is unique on initialization. Empty queries are an exception to
this, as they do not return any item in the first place.
That is because tracking how many iterations have already passed would
require regressing either normal query iteration a slight bit, or sorted
iteration by a lot. Besides, that would not be the intended use of these
methods.
## Testing
To ensure that `next` being called before `sort` results in a panic, I
added some tests. I also test that empty `QueryIter`s do not exhibit
this restriction.
The query sorts test checks for equivalence to the underlying sorts.
This change requires that `Query<(Entity, Entity)>` remains legal, if
that is not already guaranteed, which is also ensured by the
aforementioned test.
## Next Steps
Implement the set of sort methods for `QueryManyIter` as well.
- This will mostly work the same, other than needing to return a new
`QuerySortedManyIter` to account for iteration
over lists of entities that are not guaranteed to be unique. This new
query iterator will need a bit of internal restructuring
to allow for double-ended mutable iteration, while not regressing
read-only iteration.
The implementations for each pair of
- `sort`, `sort_unstable`,
- `sort_by`, sort_unstable_by,
- `sort_by_key,` `sort_by_cached_key`
are the same aside from the panic message and the sort call, so they
could be merged with an inner function.
That would require the use of higher-ranked trait bounds on
`WorldQuery::Item<'1>`, and is unclear to me whether it is currently
doable.
Iteration in QuerySortedIter might have space for improvement.
When sorting by `Entity`, an `(Entity, Entity)` lens `QueryData` is
constructed, is that worth remedying?
When table sorts are implemented, a fast path could be introduced to
these sort methods.
## Future Possibilities
Implementing `Ord` for EntityLocation might be useful.
Some papercuts in ergonomics can be improved by future Rust features:
- The additional generic parameter aside from the query lens can be
removed once this feature is stable:
`Fn -> impl Trait` (`impl Trait` in `Fn` trait return position)
- With type parameter defaults, the query lens generic can be defaulted
to `QueryData::Item`, allowing the sort methods
to look and behave like `slice::sort` when no query lens is specified.
- With TAIT, the iterator generic on `QuerySortedIter` and thus the huge
visible `impl Iterator` type in the sort function
signatures can be removed.
- With specialization, the bound on `L` could be relaxed to `QueryData`
when the underlying iterator is mutable.
## Changelog
Added `sort`, `sort_unstable`, `sort_by`, `sort_unstable_by`,
`sort_by_key`, `sort_by_cached_key` to `QueryIter`.
# Objective
- Fixes#12377
## Solution
Added simple `#[diagnostic::on_unimplemented(...)]` attributes to some
critical public traits providing a more approachable initial error
message. Where appropriate, a `note` is added indicating that a `derive`
macro is available.
## Examples
<details>
<summary>Examples hidden for brevity</summary>
Below is a collection of examples showing the new error messages
produced by this change. In general, messages will start with a more
Bevy-centric error message (e.g., _`MyComponent` is not a `Component`_),
and a note directing the user to an available derive macro where
appropriate.
### Missing `#[derive(Resource)]`
<details>
<summary>Example Code</summary>
```rust
use bevy::prelude::*;
struct MyResource;
fn main() {
App::new()
.insert_resource(MyResource)
.run();
}
```
</details>
<details>
<summary>Error Generated</summary>
```error
error[E0277]: `MyResource` is not a `Resource`
--> examples/app/empty.rs:7:26
|
7 | .insert_resource(MyResource)
| --------------- ^^^^^^^^^^ invalid `Resource`
| |
| required by a bound introduced by this call
|
= help: the trait `Resource` is not implemented for `MyResource`
= note: consider annotating `MyResource` with `#[derive(Resource)]`
= help: the following other types implement trait `Resource`:
AccessibilityRequested
ManageAccessibilityUpdates
bevy::bevy_a11y::Focus
DiagnosticsStore
FrameCount
bevy::prelude::State<S>
SystemInfo
bevy::prelude::Axis<T>
and 141 others
note: required by a bound in `bevy::prelude::App::insert_resource`
--> C:\Users\Zac\Documents\GitHub\bevy\crates\bevy_app\src\app.rs:419:31
|
419 | pub fn insert_resource<R: Resource>(&mut self, resource: R) -> &mut Self {
| ^^^^^^^^ required by this bound in `App::insert_resource`
```
</details>
### Putting A `QueryData` in a `QueryFilter` Slot
<details>
<summary>Example Code</summary>
```rust
use bevy::prelude::*;
#[derive(Component)]
struct A;
#[derive(Component)]
struct B;
fn my_system(_query: Query<&A, &B>) {}
fn main() {
App::new()
.add_systems(Update, my_system)
.run();
}
```
</details>
<details>
<summary>Error Generated</summary>
```error
error[E0277]: `&B` is not a valid `Query` filter
--> examples/app/empty.rs:9:22
|
9 | fn my_system(_query: Query<&A, &B>) {}
| ^^^^^^^^^^^^^ invalid `Query` filter
|
= help: the trait `QueryFilter` is not implemented for `&B`
= help: the following other types implement trait `QueryFilter`:
With<T>
Without<T>
bevy::prelude::Or<()>
bevy::prelude::Or<(F0,)>
bevy::prelude::Or<(F0, F1)>
bevy::prelude::Or<(F0, F1, F2)>
bevy::prelude::Or<(F0, F1, F2, F3)>
bevy::prelude::Or<(F0, F1, F2, F3, F4)>
and 28 others
note: required by a bound in `bevy::prelude::Query`
--> C:\Users\Zac\Documents\GitHub\bevy\crates\bevy_ecs\src\system\query.rs:349:51
|
349 | pub struct Query<'world, 'state, D: QueryData, F: QueryFilter = ()> {
| ^^^^^^^^^^^ required by this bound in `Query`
```
</details>
### Missing `#[derive(Component)]`
<details>
<summary>Example Code</summary>
```rust
use bevy::prelude::*;
struct A;
fn my_system(mut commands: Commands) {
commands.spawn(A);
}
fn main() {
App::new()
.add_systems(Startup, my_system)
.run();
}
```
</details>
<details>
<summary>Error Generated</summary>
```error
error[E0277]: `A` is not a `Bundle`
--> examples/app/empty.rs:6:20
|
6 | commands.spawn(A);
| ----- ^ invalid `Bundle`
| |
| required by a bound introduced by this call
|
= help: the trait `bevy::prelude::Component` is not implemented for `A`, which is required by `A: Bundle`
= note: consider annotating `A` with `#[derive(Component)]` or `#[derive(Bundle)]`
= help: the following other types implement trait `Bundle`:
TransformBundle
SceneBundle
DynamicSceneBundle
AudioSourceBundle<Source>
SpriteBundle
SpriteSheetBundle
Text2dBundle
MaterialMesh2dBundle<M>
and 34 others
= note: required for `A` to implement `Bundle`
note: required by a bound in `bevy::prelude::Commands::<'w, 's>::spawn`
--> C:\Users\Zac\Documents\GitHub\bevy\crates\bevy_ecs\src\system\commands\mod.rs:243:21
|
243 | pub fn spawn<T: Bundle>(&mut self, bundle: T) -> EntityCommands {
| ^^^^^^ required by this bound in `Commands::<'w, 's>::spawn`
```
</details>
### Missing `#[derive(Asset)]`
<details>
<summary>Example Code</summary>
```rust
use bevy::prelude::*;
struct A;
fn main() {
App::new()
.init_asset::<A>()
.run();
}
```
</details>
<details>
<summary>Error Generated</summary>
```error
error[E0277]: `A` is not an `Asset`
--> examples/app/empty.rs:7:23
|
7 | .init_asset::<A>()
| ---------- ^ invalid `Asset`
| |
| required by a bound introduced by this call
|
= help: the trait `Asset` is not implemented for `A`
= note: consider annotating `A` with `#[derive(Asset)]`
= help: the following other types implement trait `Asset`:
Font
AnimationGraph
DynamicScene
Scene
AudioSource
Pitch
bevy::bevy_gltf::Gltf
GltfNode
and 17 others
note: required by a bound in `init_asset`
--> C:\Users\Zac\Documents\GitHub\bevy\crates\bevy_asset\src\lib.rs:307:22
|
307 | fn init_asset<A: Asset>(&mut self) -> &mut Self;
| ^^^^^ required by this bound in `AssetApp::init_asset`
```
</details>
### Mismatched Input and Output on System Piping
<details>
<summary>Example Code</summary>
```rust
use bevy::prelude::*;
fn producer() -> u32 {
123
}
fn consumer(_: In<u16>) {}
fn main() {
App::new()
.add_systems(Update, producer.pipe(consumer))
.run();
}
```
</details>
<details>
<summary>Error Generated</summary>
```error
error[E0277]: `fn(bevy::prelude::In<u16>) {consumer}` is not a valid system with input `u32` and output `_`
--> examples/app/empty.rs:11:44
|
11 | .add_systems(Update, producer.pipe(consumer))
| ---- ^^^^^^^^ invalid system
| |
| required by a bound introduced by this call
|
= help: the trait `bevy::prelude::IntoSystem<u32, _, _>` is not implemented for fn item `fn(bevy::prelude::In<u16>) {consumer}`
= note: expecting a system which consumes `u32` and produces `_`
note: required by a bound in `pipe`
--> C:\Users\Zac\Documents\GitHub\bevy\crates\bevy_ecs\src\system\mod.rs:168:12
|
166 | fn pipe<B, Final, MarkerB>(self, system: B) -> PipeSystem<Self::System, B::System>
| ---- required by a bound in this associated function
167 | where
168 | B: IntoSystem<Out, Final, MarkerB>,
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ required by this bound in `IntoSystem::pipe`
```
</details>
### Missing Reflection
<details>
<summary>Example Code</summary>
```rust
use bevy::prelude::*;
#[derive(Component)]
struct MyComponent;
fn main() {
App::new()
.register_type::<MyComponent>()
.run();
}
```
</details>
<details>
<summary>Error Generated</summary>
```error
error[E0277]: `MyComponent` does not provide type registration information
--> examples/app/empty.rs:8:26
|
8 | .register_type::<MyComponent>()
| ------------- ^^^^^^^^^^^ the trait `GetTypeRegistration` is not implemented for `MyComponent`
| |
| required by a bound introduced by this call
|
= note: consider annotating `MyComponent` with `#[derive(Reflect)]`
= help: the following other types implement trait `GetTypeRegistration`:
bool
char
isize
i8
i16
i32
i64
i128
and 443 others
note: required by a bound in `bevy::prelude::App::register_type`
--> C:\Users\Zac\Documents\GitHub\bevy\crates\bevy_app\src\app.rs:619:29
|
619 | pub fn register_type<T: bevy_reflect::GetTypeRegistration>(&mut self) -> &mut Self {
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ required by this bound in `App::register_type`
```
</details>
### Missing `#[derive(States)]` Implementation
<details>
<summary>Example Code</summary>
```rust
use bevy::prelude::*;
#[derive(Debug, Clone, Copy, Default, Eq, PartialEq, Hash)]
enum AppState {
#[default]
Menu,
InGame {
paused: bool,
turbo: bool,
},
}
fn main() {
App::new()
.init_state::<AppState>()
.run();
}
```
</details>
<details>
<summary>Error Generated</summary>
```error
error[E0277]: the trait bound `AppState: FreelyMutableState` is not satisfied
--> examples/app/empty.rs:15:23
|
15 | .init_state::<AppState>()
| ---------- ^^^^^^^^ the trait `FreelyMutableState` is not implemented for `AppState`
| |
| required by a bound introduced by this call
|
= note: consider annotating `AppState` with `#[derive(States)]`
note: required by a bound in `bevy::prelude::App::init_state`
--> C:\Users\Zac\Documents\GitHub\bevy\crates\bevy_app\src\app.rs:282:26
|
282 | pub fn init_state<S: FreelyMutableState + FromWorld>(&mut self) -> &mut Self {
| ^^^^^^^^^^^^^^^^^^ required by this bound in `App::init_state`
```
</details>
### Adding a `System` with Unhandled Output
<details>
<summary>Example Code</summary>
```rust
use bevy::prelude::*;
fn producer() -> u32 {
123
}
fn main() {
App::new()
.add_systems(Update, consumer)
.run();
}
```
</details>
<details>
<summary>Error Generated</summary>
```error
error[E0277]: `fn() -> u32 {producer}` does not describe a valid system configuration
--> examples/app/empty.rs:9:30
|
9 | .add_systems(Update, producer)
| ----------- ^^^^^^^^ invalid system configuration
| |
| required by a bound introduced by this call
|
= help: the trait `IntoSystem<(), (), _>` is not implemented for fn item `fn() -> u32 {producer}`, which is required by `fn() -> u32 {producer}: IntoSystemConfigs<_>`
= help: the following other types implement trait `IntoSystemConfigs<Marker>`:
<Box<(dyn bevy::prelude::System<In = (), Out = ()> + 'static)> as IntoSystemConfigs<()>>
<NodeConfigs<Box<(dyn bevy::prelude::System<In = (), Out = ()> + 'static)>> as IntoSystemConfigs<()>>
<(S0,) as IntoSystemConfigs<(SystemConfigTupleMarker, P0)>>
<(S0, S1) as IntoSystemConfigs<(SystemConfigTupleMarker, P0, P1)>>
<(S0, S1, S2) as IntoSystemConfigs<(SystemConfigTupleMarker, P0, P1, P2)>>
<(S0, S1, S2, S3) as IntoSystemConfigs<(SystemConfigTupleMarker, P0, P1, P2, P3)>>
<(S0, S1, S2, S3, S4) as IntoSystemConfigs<(SystemConfigTupleMarker, P0, P1, P2, P3, P4)>>
<(S0, S1, S2, S3, S4, S5) as IntoSystemConfigs<(SystemConfigTupleMarker, P0, P1, P2, P3, P4, P5)>>
and 14 others
= note: required for `fn() -> u32 {producer}` to implement `IntoSystemConfigs<_>`
note: required by a bound in `bevy::prelude::App::add_systems`
--> C:\Users\Zac\Documents\GitHub\bevy\crates\bevy_app\src\app.rs:342:23
|
339 | pub fn add_systems<M>(
| ----------- required by a bound in this associated function
...
342 | systems: impl IntoSystemConfigs<M>,
| ^^^^^^^^^^^^^^^^^^^^ required by this bound in `App::add_systems`
```
</details>
</details>
## Testing
CI passed locally.
## Migration Guide
Upgrade to version 1.78 (or higher) of Rust.
## Future Work
- Currently, hints are not supported in this diagnostic. Ideally,
suggestions like _"consider using ..."_ would be in a hint rather than a
note, but that is the best option for now.
- System chaining and other `all_tuples!(...)`-based traits have bad
error messages due to the slightly different error message format.
---------
Co-authored-by: Jamie Ridding <Themayu@users.noreply.github.com>
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
Co-authored-by: BD103 <59022059+BD103@users.noreply.github.com>
# Objective
Provides a `WorldQuery` implementation on `Mut<T>` that forwards to the
implementation on `&mut T`, and give users a way to opt-in to change
detection in auto-generated `QueryData::ReadOnly` types.
Fixes#13329.
## Solution
I implemented `WorldQuery` on `Mut<'w, T>` as a forwarding
implementation to `&mut T`, setting the `QueryData::ReadOnly` associated
type to `Ref<'w, T>`. This provides users the ability to explicitly
opt-in to change detection in the read-only forms of queries.
## Testing
A documentation test was added to `Mut` showcasing the new
functionality.
---
## Changelog
### Added
- Added an implementation of `WorldQuery` and `QueryData` on
`bevy_ecs::change_detection::Mut`.
# Objective
Passing `&World` in the `WorldQuery::get_state` method is unnecessary,
as all implementations of this method in the engine either only access
`Components` in `&World`, or do nothing with it.
It can introduce UB by necessitating the creation of a `&World` from a
`UnsafeWorldCell`.
This currently happens in `Query::transmute_lens`, which obtains a
`&World` from the internal `UnsafeWorldCell` solely to pass to
`get_state`. `Query::join` suffers from the same issue.
Other cases of UB come from allowing implementors of `WorldQuery` to
freely access `&World`, like in the `bevy-trait-query` crate, where a
[reference to a resource is
obtained](0c0e7dd646/src/lib.rs (L445))
inside of
[`get_state`](0c0e7dd646/src/one.rs (L245)),
potentially aliasing with a `ResMut` parameter in the same system.
`WorldQuery::init_state` currently requires `&mut World`, which doesn't
suffer from these issues.
But that too can be changed to receive a wrapper around `&mut
Components` and `&mut Storages` for consistency in a follow-up PR.
## Solution
Replace the `&World` parameter in `get_state` with `&Components`.
## Changelog
`WorldQuery::get_state` now takes `&Components` instead of `&World`.
The `transmute`, `transmute_filtered`, `join` and `join_filtered`
methods on `QueryState` now similarly take `&Components` instead of
`&World`.
## Migration Guide
Users of `WorldQuery::get_state` or `transmute`, `transmute_filtered`,
`join` and `join_filtered` methods on `QueryState` now need to pass
`&Components` instead of `&World`.
`&Components` can be trivially obtained from either `components` method
on `&World` or `UnsafeWorldCell`.
For implementors of `WorldQuery::get_state` that were accessing more
than the `Components` inside `&World` and its methods, this is no longer
allowed.
# Objective
Fixes#12966
## Solution
Renaming multi_threaded feature to match snake case
## Migration Guide
Bevy feature multi-threaded should be refered to multi_threaded from now
on.
# Objective
- bevy usually use `Parallel::scope` to collect items from `par_iter`,
but `scope` will be called with every satifified items. it will cause a
lot of unnecessary lookup.
## Solution
- similar to Rayon ,we introduce `for_each_init` for `par_iter` which
only be invoked when spawn a task for a group of items.
---
## Changelog
- added `for_each_init`
## Performance
`check_visibility ` in `many_foxes `
~40% performance gain in `check_visibility`.
---------
Co-authored-by: James Liu <contact@jamessliu.com>
# Objective
Allow parallel iteration over events, resolve#10766
## Solution
- Add `EventParIter` which works similarly to `QueryParIter`,
implementing a `for_each{_with_id}` operator.
I chose to not mirror `EventIteratorWithId` and instead implement both
operations on a single struct.
- Reuse `BatchingStrategy` from `QueryParIter`
## Changelog
- `EventReader` now supports parallel event iteration using
`par_read().for_each(|event| ...)`.
---------
Co-authored-by: James Liu <contact@jamessliu.com>
Co-authored-by: Pablo Reinhardt <126117294+pablo-lua@users.noreply.github.com>
# Objective
Fix#2128. Both `Query::new_archetype` and `SystemParam::new_archetype`
do not check if the `Archetype` comes from the same World the state is
initialized from. This could result in unsoundness via invalid accesses
if called incorrectly.
## Solution
Make them `unsafe` functions and lift the invariant to the caller. This
also caught one instance of us not validating the World in
`SystemState::update_archetypes_unsafe_world_cell`'s implementation.
---
## Changelog
Changed: `QueryState::new_archetype` is now an unsafe function.
Changed: `SystemParam::new_archetype` is now an unsafe function.
## Migration Guide
`QueryState::new_archetype` and `SystemParam::new_archetype` are now an
unsafe functions that must be sure that the provided `Archetype` is from
the same `World` that the state was initialized from. Callers may need
to add additional assertions or propagate the safety invariant upwards
through the callstack to ensure safety.
# Objective
- I daily drive nightly Rust when developing Bevy, so I notice when new
warnings are raised by `cargo check` and Clippy.
- `cargo +nightly clippy` raises a few of these new warnings.
## Solution
- Fix most warnings from `cargo +nightly clippy`
- I skipped the docs-related warnings because some were covered by
#12692.
- Use `Clone::clone_from` in applicable scenarios, which can sometimes
avoid an extra allocation.
- Implement `Default` for structs that have a `pub const fn new() ->
Self` method.
- Fix an occurrence where generic constraints were defined in both `<C:
Trait>` and `where C: Trait`.
- Removed generic constraints that were implied by the `Bundle` trait.
---
## Changelog
- `BatchingStrategy`, `NonGenericTypeCell`, and `GenericTypeCell` now
implement `Default`.
# Objective
- Fix#7303
- bevy would spawn a lot of tasks in parallel iteration when it matchs a
large storage and many small storage ,it significantly increase the
overhead of schedule.
## Solution
- collect small storage into one task
# Objective
- There are several redundant imports in the tests and examples that are
not caught by CI because additional flags need to be passed.
## Solution
- Run `cargo check --workspace --tests` and `cargo check --workspace
--examples`, then fix all warnings.
- Add `test-check` to CI, which will be run in the check-compiles job.
This should catch future warnings for tests. Examples are already
checked, but I'm not yet sure why they weren't caught.
## Discussion
- Should the `--tests` and `--examples` flags be added to CI, so this is
caught in the future?
- If so, #12818 will need to be merged first. It was also a warning
raised by checking the examples, but I chose to split off into a
separate PR.
---------
Co-authored-by: François Mockers <francois.mockers@vleue.com>
# Objective
Other than the exposed functions for reading matched tables and
archetypes, a `QueryState` does not actually need both internal Vecs for
storing matched archetypes and tables. In practice, it will only use one
of the two depending on if it uses dense or archetypal iteration.
Same vein as #12474. The goal is to reduce the memory overhead of using
queries, which Bevy itself, ecosystem plugins, and end users are already
fairly liberally using.
## Solution
Add `StorageId`, which is a union over `TableId` and `ArchetypeId`, and
store only one of the two at runtime. Read the slice as if it was one ID
depending on whether the query is dense or not.
This follows in the same vein as #5085; however, this one directly
impacts heap memory usage at runtime, while #5085 primarily targeted
transient pointers that might not actually exist at runtime.
---
## Changelog
Changed: `QueryState::matched_tables` now returns an iterator instead of
a reference to a slice.
Changed: `QueryState::matched_archetypes` now returns an iterator
instead of a reference to a slice.
## Migration Guide
`QueryState::matched_tables` and `QueryState::matched_archetypes` does
not return a reference to a slice, but an iterator instead. You may need
to use iterator combinators or collect them into a Vec to use it as a
slice.
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
# Objective
Fixes#12752. Fixes#12750. Document the runtime complexity of all of
the `O(1)` operations on the individual APIs.
## Solution
* Mirror `Query::contains` onto `QueryState::contains`
* Make `QueryState::as_nop` pub(crate)
* Make `NopWorldQuery` pub(crate)
* Document all of the O(1) operations on Query and QueryState.
# Objective
Fixes#12392, fixes#12393, and fixes#11387. Implement QueryData for
Archetype and EntityLocation.
## Solution
Add impls for both of the types.
---
## Changelog
Added: `&Archetype` now implements `QueryData`
Added: `EntityLocation` now implements `QueryData`
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
# Objective
I'm reading through the ecs query code for the first time, and updating
the docs:
- fixed some typos
- added some docs about things I was confused about (in particular what
the difference between `matches_component_set` and
`update_component_access` was)
# Objective
`QueryState::archetype_component_access` is only really ever used to
extend `SystemMeta`'s. It can be removed to save some memory for every
`Query` in an app.
## Solution
* Remove it.
* Have `new_archetype` pass in a `&mut Access<ArchetypeComponentId>`
instead and pull it from `SystemMeta` directly.
* Split `QueryState::new` from `QueryState::new_with_access` and a
common `QueryState::new_uninitialized`.
* Split `new_archetype` into an internal and public version. Call the
internal version in `update_archetypes`.
This should make it faster to construct new QueryStates, and by proxy
lenses and joins as well.
`matched_tables` also similarly is only used to deduplicate inserting
into `matched_table_ids`. If we can find another efficient way to do so,
it might also be worth removing.
The [generated
assembly](https://github.com/james7132/bevy_asm_tests/compare/main...remove-query-state-archetype-component-access#diff-496530101f0b16e495b7e9b77c0e906ae3068c8adb69ed36c92d5a1be5a9efbe)
reflects this well, with all of the access related updates in
`QueryState` being removed.
---
## Changelog
Removed: `QueryState::archetype_component_access`.
Changed: `QueryState::new_archetype` now takes a `&mut
Access<ArchetypeComponentId>` argument, which will be updated with the
new accesses.
Changed: `QueryState::update_archetype_component_access` now takes a
`&mut Access<ArchetypeComponentId>` argument, which will be updated with
the new accesses.
## Migration Guide
TODO
# Objective
Improve code quality involving fixedbitset.
## Solution
Update to fixedbitset 0.5. Use the new `grow_and_insert` function
instead of `grow` and `insert` functions separately.
This should also speed up most of the set operations involving
fixedbitset. They should be ~2x faster, but testing this against the
stress tests seems to show little to no difference. The multithreaded
executor doesn't seem to be all that much faster in many_cubes and
many_foxes. These use cases are likely dominated by other operations or
the bitsets aren't big enough to make them the bottleneck.
This introduces a duplicate dependency due to petgraph and wgpu, but the
former may take some time to update.
## Changelog
Removed: `Access::grow`
## Migration Guide
`Access::grow` has been removed. It's no longer needed. Remove all
references to it.
# Objective
- Add a way to combine 2 queries together in a similar way to
`Query::transmute_lens`
- Fixes#1658
## Solution
- Use a similar method to query transmute, but take the intersection of
matched archetypes between the 2 queries and the union of the accesses
to create the new underlying QueryState.
---
## Changelog
- Add query joins
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
# Objective
Fix#10876. Improve `Query` and `QueryState`'s docs.
## Solution
Explicitly denote that Query is always guaranteed to return results from
all matching entities once and only once for each entity, and that
iteration order is not guaranteed in any way.
# Objective
When doing a final pass for #3362, it appeared that `ComponentStorage`
as a trait, the two types implementing it, and the associated type on
`Component` aren't really necessary anymore. This likely was due to an
earlier constraint on the use of consts in traits, but that definitely
doesn't seem to be a problem in Rust 1.76.
## Solution
Remove them.
---
## Changelog
Changed: `Component::Storage` has been replaced with
`Component::STORAGE_TYPE` as a const.
Removed: `bevy::ecs::component::ComponentStorage` trait
Removed: `bevy::ecs::component::TableStorage` struct
Removed: `bevy::ecs::component::SparseSetStorage` struct
## Migration Guide
If you were manually implementing `Component` instead of using the
derive macro, replace the associated `Storage` associated type with the
`STORAGE_TYPE` const:
```rust
// in Bevy 0.13
impl Component for MyComponent {
type Storage = TableStorage;
}
// in Bevy 0.14
impl Component for MyComponent {
const STORAGE_TYPE: StorageType = StorageType::Table;
}
```
Component is no longer object safe. If you were relying on `&dyn
Component`, `Box<dyn Component>`, etc. please [file an issue
](https://github.com/bevyengine/bevy/issues) to get [this
change](https://github.com/bevyengine/bevy/pull/12311) reverted.
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
# Objective
- Add the new `-Zcheck-cfg` checks to catch more warnings
- Fixes#12091
## Solution
- Create a new `cfg-check` to the CI that runs `cargo check -Zcheck-cfg
--workspace` using cargo nightly (and fails if there are warnings)
- Fix all warnings generated by the new check
---
## Changelog
- Remove all redundant imports
- Fix cfg wasm32 targets
- Add 3 dead code exceptions (should StandardColor be unused?)
- Convert ios_simulator to a feature (I'm not sure if this is the right
way to do it, but the check complained before)
## Migration Guide
No breaking changes
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
# Objective
- Part of #11590
- Fix `unsafe_op_in_unsafe_fn` for trivial cases in bevy_ecs
## Solution
Fix `unsafe_op_in_unsafe_fn` in bevy_ecs for trivial cases, i.e., add an
`unsafe` block when the safety comment already exists or add a comment
like "The invariants are uphold by the caller".
---------
Co-authored-by: James Liu <contact@jamessliu.com>
# Objective
We deprecated quite a few APIs in 0.13. 0.13 has shipped already. It
should be OK to remove them in 0.14's release. Fixes#4059. Fixes#9011.
## Solution
Remove them.
# Objective
`update_archetype_component_access` was removed from queries in #9774,
but some documentation still refers to it.
## Solution
Update the documentation. Since a bunch of these were in SAFETY comments
it would be nice if someone who knows the details better could check
that the rest of those comments are still valid.
# Objective
It would be useful to be able to inspect a `QueryState`'s accesses so we
can detect when the data it accesses changes without having to iterate
it. However there are two things preventing this:
* These accesses are unnecessarily encapsulated.
* `Has<T>` indirectly accesses `T`, but does not register it.
## Solution
* Expose accesses and matches used by `QueryState`.
* Add the notion of "archetypal" accesses, which are not accessed
directly, but whose presence in an archetype affects a query result.
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
# Objective
- (Partially) Fixes#9904
- Acts on #9910
## Solution
- Deprecated the relevant methods from `Query`, cascading changes as
required across Bevy.
---
## Changelog
- Deprecated `QueryState::get_component_unchecked_mut` method
- Deprecated `Query::get_component` method
- Deprecated `Query::get_component_mut` method
- Deprecated `Query::component` method
- Deprecated `Query::component_mut` method
- Deprecated `Query::get_component_unchecked_mut` method
## Migration Guide
### `QueryState::get_component_unchecked_mut`
Use `QueryState::get_unchecked_manual` and select for the exact
component based on the structure of the exact query as required.
### `Query::(get_)component(_unchecked)(_mut)`
Use `Query::get` and select for the exact component based on the
structure of the exact query as required.
- For mutable access (`_mut`), use `Query::get_mut`
- For unchecked access (`_unchecked`), use `Query::get_unchecked`
- For panic variants (non-`get_`), add `.unwrap()`
## Notes
- `QueryComponentError` can be removed once these deprecated methods are
also removed. Due to an interaction with `thiserror`'s derive macro, it
is not marked as deprecated.
# Objective
Fixes#11311
## Solution
Adds an example to the documentation for `par_iter_mut`. I didn't add
any examples to `par_iter`, because I couldn't think of a good example
and I figure users can infer that `par_iter` and `par_iter_mut` are
similar.
# Objective
Expand the existing `Query` API to support more dynamic use cases i.e.
scripting.
## Prior Art
- #6390
- #8308
- #10037
## Solution
- Create a `QueryBuilder` with runtime methods to define the set of
component accesses for a built query.
- Create new `WorldQueryData` implementations `FilteredEntityMut` and
`FilteredEntityRef` as variants of `EntityMut` and `EntityRef` that
provide run time checked access to the components included in a given
query.
- Add new methods to `Query` to create "query lens" with a subset of the
access of the initial query.
### Query Builder
The `QueryBuilder` API allows you to define a query at runtime. At it's
most basic use it will simply create a query with the corresponding type
signature:
```rust
let query = QueryBuilder::<Entity, With<A>>::new(&mut world).build();
// is equivalent to
let query = QueryState::<Entity, With<A>>::new(&mut world);
```
Before calling `.build()` you also have the opportunity to add
additional accesses and filters. Here is a simple example where we add
additional filter terms:
```rust
let entity_a = world.spawn((A(0), B(0))).id();
let entity_b = world.spawn((A(0), C(0))).id();
let mut query_a = QueryBuilder::<Entity>::new(&mut world)
.with::<A>()
.without::<C>()
.build();
assert_eq!(entity_a, query_a.single(&world));
```
This alone is useful in that allows you to decide which archetypes your
query will match at runtime. However it is also very limited, consider a
case like the following:
```rust
let query_a = QueryBuilder::<&A>::new(&mut world)
// Add an additional access
.data::<&B>()
.build();
```
This will grant the query an additional read access to component B
however we have no way of accessing the data while iterating as the type
signature still only includes &A. For an even more concrete example of
this consider dynamic components:
```rust
let query_a = QueryBuilder::<Entity>::new(&mut world)
// Adding a filter is easy since it doesn't need be read later
.with_id(component_id_a)
// How do I access the data of this component?
.ref_id(component_id_b)
.build();
```
With this in mind the `QueryBuilder` API seems somewhat incomplete by
itself, we need some way method of accessing the components dynamically.
So here's one:
### Query Transmutation
If the problem is not having the component in the type signature why not
just add it? This PR also adds transmute methods to `QueryBuilder` and
`QueryState`. Here's a simple example:
```rust
world.spawn(A(0));
world.spawn((A(1), B(0)));
let mut query = QueryBuilder::<()>::new(&mut world)
.with::<B>()
.transmute::<&A>()
.build();
query.iter(&world).for_each(|a| assert_eq!(a.0, 1));
```
The `QueryState` and `QueryBuilder` transmute methods look quite similar
but are different in one respect. Transmuting a builder will always
succeed as it will just add the additional accesses needed for the new
terms if they weren't already included. Transmuting a `QueryState` will
panic in the case that the new type signature would give it access it
didn't already have, for example:
```rust
let query = QueryState::<&A, Option<&B>>::new(&mut world);
/// This is fine, the access for Option<&A> is less restrictive than &A
query.transmute::<Option<&A>>(&world);
/// Oh no, this would allow access to &B on entities that might not have it, so it panics
query.transmute::<&B>(&world);
/// This is right out
query.transmute::<&C>(&world);
```
This is quite an appealing API to also have available on `Query` however
it does pose one additional wrinkle: In order to to change the iterator
we need to create a new `QueryState` to back it. `Query` doesn't own
it's own state though, it just borrows it, so we need a place to borrow
it from. This is why `QueryLens` exists, it is a place to store the new
state so it can be borrowed when you call `.query()` leaving you with an
API like this:
```rust
fn function_that_takes_a_query(query: &Query<&A>) {
// ...
}
fn system(query: Query<(&A, &B)>) {
let lens = query.transmute_lens::<&A>();
let q = lens.query();
function_that_takes_a_query(&q);
}
```
Now you may be thinking: Hey, wait a second, you introduced the problem
with dynamic components and then described a solution that only works
for static components! Ok, you got me, I guess we need a bit more:
### Filtered Entity References
Currently the only way you can access dynamic components on entities
through a query is with either `EntityMut` or `EntityRef`, however these
can access all components and so conflict with all other accesses. This
PR introduces `FilteredEntityMut` and `FilteredEntityRef` as
alternatives that have additional runtime checking to prevent accessing
components that you shouldn't. This way you can build a query with a
`QueryBuilder` and actually access the components you asked for:
```rust
let mut query = QueryBuilder::<FilteredEntityRef>::new(&mut world)
.ref_id(component_id_a)
.with(component_id_b)
.build();
let entity_ref = query.single(&world);
// Returns Some(Ptr) as we have that component and are allowed to read it
let a = entity_ref.get_by_id(component_id_a);
// Will return None even though the entity does have the component, as we are not allowed to read it
let b = entity_ref.get_by_id(component_id_b);
```
For the most part these new structs have the exact same methods as their
non-filtered equivalents.
Putting all of this together we can do some truly dynamic ECS queries,
check out the `dynamic` example to see it in action:
```
Commands:
comp, c Create new components
spawn, s Spawn entities
query, q Query for entities
Enter a command with no parameters for usage.
> c A, B, C, Data 4
Component A created with id: 0
Component B created with id: 1
Component C created with id: 2
Component Data created with id: 3
> s A, B, Data 1
Entity spawned with id: 0v0
> s A, C, Data 0
Entity spawned with id: 1v0
> q &Data
0v0: Data: [1, 0, 0, 0]
1v0: Data: [0, 0, 0, 0]
> q B, &mut Data
0v0: Data: [2, 1, 1, 1]
> q B || C, &Data
0v0: Data: [2, 1, 1, 1]
1v0: Data: [0, 0, 0, 0]
```
## Changelog
- Add new `transmute_lens` methods to `Query`.
- Add new types `QueryBuilder`, `FilteredEntityMut`, `FilteredEntityRef`
and `QueryLens`
- `update_archetype_component_access` has been removed, archetype
component accesses are now determined by the accesses set in
`update_component_access`
- Added method `set_access` to `WorldQuery`, this is called before
`update_component_access` for queries that have a restricted set of
accesses, such as those built by `QueryBuilder` or `QueryLens`. This is
primarily used by the `FilteredEntity*` variants and has an empty trait
implementation.
- Added method `get_state` to `WorldQuery` as a fallible version of
`init_state` when you don't have `&mut World` access.
## Future Work
Improve performance of `FilteredEntityMut` and `FilteredEntityRef`,
currently they have to determine the accesses a query has in a given
archetype during iteration which is far from ideal, especially since we
already did the work when matching the archetype in the first place. To
avoid making more internal API changes I have left it out of this PR.
---------
Co-authored-by: Mike Hsu <mike.hsu@gmail.com>
# Objective
There are a lot of doctests that are `ignore`d for no documented reason.
And that should be fixed.
## Solution
I searched the bevy repo with the regex ` ```[a-z,]*ignore ` in order to
find all `ignore`d doctests. For each one of the `ignore`d doctests, I
did the following steps:
1. Attempt to remove the `ignored` attribute while still passing the
test. I did this by adding hidden dummy structs and imports.
2. If step 1 doesn't work, attempt to replace the `ignored` attribute
with the `no_run` attribute while still passing the test.
3. If step 2 doesn't work, keep the `ignored` attribute but add
documentation for why the `ignored` attribute was added.
---------
Co-authored-by: François <mockersf@gmail.com>
# Objective
Fixes#11050
Rename ArchetypeEntity::entity to ArchetypeEntity::id to be consistent
with `EntityWorldMut`, `EntityMut` and `EntityRef`.
## Migration Guide
The method `ArchetypeEntity::entity` has been renamed to
`ArchetypeEntity::id`
# Objective
Fix ci hang, so we can merge pr's again.
## Solution
- switch ppa action to use mesa stable versions
https://launchpad.net/~kisak/+archive/ubuntu/turtle
- use commit from #11123
---------
Co-authored-by: Stepan Koltsov <stepan.koltsov@gmail.com>
# Objective
`Has<T>` in some niche cases may behave in an unexpected way.
Specifically, when using `Query::get` on a `Has<T>` with a despawned
entity.
## Solution
Add precision about cases wehre `Query::get` could return an `Err`.
# Objective
The definition of several `QueryState` methods use unnecessary explicit
lifetimes, which adds to visual noise.
## Solution
Elide the lifetimes.
