# Objective
Fixes#15541
A bunch of lifetimes were added during the Assets V2 rework, but after
moving to async traits in #12550 they can be elided. That PR mentions
that this might be the case, but apparently it wasn't followed up on at
the time.
~~I ended up grepping for `<'a` and finding a similar case in
`bevy_reflect` which I also fixed.~~ (edit: that one was needed
apparently)
Note that elided lifetimes are unstable in `impl Trait`. If that gets
stabilized then we can elide even more.
## Solution
Remove the extra lifetimes.
## Testing
Everything still compiles. If I have messed something up there is a
small risk that some user code stops compiling, but all the examples
still work at least.
---
## Migration Guide
The traits `AssetLoader`, `AssetSaver` and `Process` traits from
`bevy_asset` now use elided lifetimes. If you implement these then
remove the named lifetime.
# Objective
This PR extends and reworks the material from #15282 by allowing
arbitrary curves to be used by the animation system to animate arbitrary
properties. The goals of this work are to:
- Allow far greater flexibility in how animations are allowed to be
defined in order to be used with `bevy_animation`.
- Delegate responsibility over keyframe interpolation to `bevy_math` and
the `Curve` libraries and reduce reliance on keyframes in animation
definitions generally.
- Move away from allowing the glTF spec to completely define animations
on a mechanical level.
## Solution
### Overview
At a high level, curves have been incorporated into the animation system
using the `AnimationCurve` trait (closely related to what was
`Keyframes`). From the top down:
1. In `animate_targets`, animations are driven by `VariableCurve`, which
is now a thin wrapper around a `Box<dyn AnimationCurve>`.
2. `AnimationCurve` is something built out of a `Curve`, and it tells
the animation system how to use the curve's output to actually mutate
component properties. The trait looks like this:
```rust
/// A low-level trait that provides control over how curves are actually applied to entities
/// by the animation system.
///
/// Typically, this will not need to be implemented manually, since it is automatically
/// implemented by [`AnimatableCurve`] and other curves used by the animation system
/// (e.g. those that animate parts of transforms or morph weights). However, this can be
/// implemented manually when `AnimatableCurve` is not sufficiently expressive.
///
/// In many respects, this behaves like a type-erased form of [`Curve`], where the output
/// type of the curve is remembered only in the components that are mutated in the
/// implementation of [`apply`].
///
/// [`apply`]: AnimationCurve::apply
pub trait AnimationCurve: Reflect + Debug + Send + Sync {
/// Returns a boxed clone of this value.
fn clone_value(&self) -> Box<dyn AnimationCurve>;
/// The range of times for which this animation is defined.
fn domain(&self) -> Interval;
/// Write the value of sampling this curve at time `t` into `transform` or `entity`,
/// as appropriate, interpolating between the existing value and the sampled value
/// using the given `weight`.
fn apply<'a>(
&self,
t: f32,
transform: Option<Mut<'a, Transform>>,
entity: EntityMutExcept<'a, (Transform, AnimationPlayer, Handle<AnimationGraph>)>,
weight: f32,
) -> Result<(), AnimationEvaluationError>;
}
```
3. The conversion process from a `Curve` to an `AnimationCurve` involves
using wrappers which communicate the intent to animate a particular
property. For example, here is `TranslationCurve`, which wraps a
`Curve<Vec3>` and uses it to animate `Transform::translation`:
```rust
/// This type allows a curve valued in `Vec3` to become an [`AnimationCurve`] that animates
/// the translation component of a transform.
pub struct TranslationCurve<C>(pub C);
```
### Animatable Properties
The `AnimatableProperty` trait survives in the transition, and it can be
used to allow curves to animate arbitrary component properties. The
updated documentation for `AnimatableProperty` explains this process:
<details>
<summary>Expand AnimatableProperty example</summary
An `AnimatableProperty` is a value on a component that Bevy can animate.
You can implement this trait on a unit struct in order to support
animating
custom components other than transforms and morph weights. Use that type
in
conjunction with `AnimatableCurve` (and perhaps
`AnimatableKeyframeCurve`
to define the animation itself). For example, in order to animate font
size of a
text section from 24 pt. to 80 pt., you might use:
```rust
#[derive(Reflect)]
struct FontSizeProperty;
impl AnimatableProperty for FontSizeProperty {
type Component = Text;
type Property = f32;
fn get_mut(component: &mut Self::Component) -> Option<&mut Self::Property> {
Some(&mut component.sections.get_mut(0)?.style.font_size)
}
}
```
You can then create an `AnimationClip` to animate this property like so:
```rust
let mut animation_clip = AnimationClip::default();
animation_clip.add_curve_to_target(
animation_target_id,
AnimatableKeyframeCurve::new(
[
(0.0, 24.0),
(1.0, 80.0),
]
)
.map(AnimatableCurve::<FontSizeProperty, _>::from_curve)
.expect("Failed to create font size curve")
);
```
Here, the use of `AnimatableKeyframeCurve` creates a curve out of the
given keyframe time-value
pairs, using the `Animatable` implementation of `f32` to interpolate
between them. The
invocation of `AnimatableCurve::from_curve` with `FontSizeProperty`
indicates that the `f32`
output from that curve is to be used to animate the font size of a
`Text` component (as
configured above).
</details>
### glTF Loading
glTF animations are now loaded into `Curve` types of various kinds,
depending on what is being animated and what interpolation mode is being
used. Those types get wrapped into and converted into `Box<dyn
AnimationCurve>` and shoved inside of a `VariableCurve` just like
everybody else.
### Morph Weights
There is an `IterableCurve` abstraction which allows sampling these from
a contiguous buffer without allocating. Its only reason for existing is
that Rust disallows you from naming function types, otherwise we would
just use `Curve` with an iterator output type. (The iterator involves
`Map`, and the name of the function type would have to be able to be
named, but it is not.)
A `WeightsCurve` adaptor turns an `IterableCurve` into an
`AnimationCurve`, so it behaves like everything else in that regard.
## Testing
Tested by running existing animation examples. Interpolation logic has
had additional tests added within the `Curve` API to replace the tests
in `bevy_animation`. Some kinds of out-of-bounds errors have become
impossible.
Performance testing on `many_foxes` (`animate_targets`) suggests that
performance is very similar to the existing implementation. Here are a
couple trace histograms across different runs (yellow is this branch,
red is main).
<img width="669" alt="Screenshot 2024-09-27 at 9 41 50 AM"
src="https://github.com/user-attachments/assets/5ba4e9ac-3aea-452e-aaf8-1492acc2d7fc">
<img width="673" alt="Screenshot 2024-09-27 at 9 45 18 AM"
src="https://github.com/user-attachments/assets/8982538b-04cf-46b5-97b2-164c6bc8162e">
---
## Migration Guide
Most user code that does not directly deal with `AnimationClip` and
`VariableCurve` will not need to be changed. On the other hand,
`VariableCurve` has been completely overhauled. If you were previously
defining animation curves in code using keyframes, you will need to
migrate that code to use curve constructors instead. For example, a
rotation animation defined using keyframes and added to an animation
clip like this:
```rust
animation_clip.add_curve_to_target(
animation_target_id,
VariableCurve {
keyframe_timestamps: vec![0.0, 1.0, 2.0, 3.0, 4.0],
keyframes: Keyframes::Rotation(vec![
Quat::IDENTITY,
Quat::from_axis_angle(Vec3::Y, PI / 2.),
Quat::from_axis_angle(Vec3::Y, PI / 2. * 2.),
Quat::from_axis_angle(Vec3::Y, PI / 2. * 3.),
Quat::IDENTITY,
]),
interpolation: Interpolation::Linear,
},
);
```
would now be added like this:
```rust
animation_clip.add_curve_to_target(
animation_target_id,
AnimatableKeyframeCurve::new([0.0, 1.0, 2.0, 3.0, 4.0].into_iter().zip([
Quat::IDENTITY,
Quat::from_axis_angle(Vec3::Y, PI / 2.),
Quat::from_axis_angle(Vec3::Y, PI / 2. * 2.),
Quat::from_axis_angle(Vec3::Y, PI / 2. * 3.),
Quat::IDENTITY,
]))
.map(RotationCurve)
.expect("Failed to build rotation curve"),
);
```
Note that the interface of `AnimationClip::add_curve_to_target` has also
changed (as this example shows, if subtly), and now takes its curve
input as an `impl AnimationCurve`. If you need to add a `VariableCurve`
directly, a new method `add_variable_curve_to_target` accommodates that
(and serves as a one-to-one migration in this regard).
### For reviewers
The diff is pretty big, and the structure of some of the changes might
not be super-obvious:
- `keyframes.rs` became `animation_curves.rs`, and `AnimationCurve` is
based heavily on `Keyframes`, with the adaptors also largely following
suite.
- The Curve API adaptor structs were moved from `bevy_math::curve::mod`
into their own module `adaptors`. There are no functional changes to how
these adaptors work; this is just to make room for the specialized
reflection implementations since `mod.rs` was getting kind of cramped.
- The new module `gltf_curves` holds the additional curve constructions
that are needed by the glTF loader. Note that the loader uses a mix of
these and off-the-shelf `bevy_math` curve stuff.
- `animatable.rs` no longer holds logic related to keyframe
interpolation, which is now delegated to the existing abstractions in
`bevy_math::curve::cores`.
---------
Co-authored-by: Gino Valente <49806985+MrGVSV@users.noreply.github.com>
Co-authored-by: aecsocket <43144841+aecsocket@users.noreply.github.com>
# Objective
Fixes#15367.
Currently, required components can only be defined through the `require`
macro attribute. While this should be used in most cases, there are also
several instances where you may want to define requirements at runtime,
commonly in plugins.
Example use cases:
- Require components only if the relevant optional plugins are enabled.
For example, a `SleepTimer` component (for physics) is only relevant if
the `SleepPlugin` is enabled.
- Third party crates can define their own requirements for first party
types. For example, "each `Handle<Mesh>` should require my custom
rendering data components". This also gets around the orphan rule.
- Generic plugins that add marker components based on the existence of
other components, like a generic `ColliderPlugin<C: AnyCollider>` that
wants to add a `ColliderMarker` component for all types of colliders.
- This is currently relevant for the retained render world in #15320.
The `ExtractComponentPlugin<C>` should add `SyncToRenderWorld` to all
components that should be extracted. This is currently done with
observers, which is more expensive than required components, and causes
archetype moves.
- Replace some built-in components with custom versions. For example, if
`GlobalTransform` required `Transform` through `TransformPlugin`, but we
wanted to use a `CustomTransform` type, we could replace
`TransformPlugin` with our own plugin. (This specific example isn't
good, but there are likely better use cases where this may be useful)
See #15367 for more in-depth reasoning.
## Solution
Add `register_required_components::<T, R>` and
`register_required_components_with::<T, R>` methods for `Default` and
custom constructors respectively. These methods exist on `App` and
`World`.
```rust
struct BirdPlugin;
impl Plugin for BirdPlugin {
fn plugin(app: &mut App) {
// Make `Bird` require `Wings` with a `Default` constructor.
app.register_required_components::<Bird, Wings>();
// Make `Wings` require `FlapSpeed` with a custom constructor.
// Fun fact: Some hummingbirds can flutter their wings 80 times per second!
app.register_required_components_with::<Wings, FlapSpeed>(|| FlapSpeed::from_duration(1.0 / 80.0));
}
}
```
The custom constructor is a function pointer to match the `require` API,
though it could take a raw value too.
Requirement inheritance works similarly as with the `require` attribute.
If `Bird` required `FlapSpeed` directly, it would take precedence over
indirectly requiring it through `Wings`. The same logic applies to all
levels of the inheritance tree.
Note that registering the same component requirement more than once will
panic, similarly to trying to add multiple component hooks of the same
type to the same component. This avoids constructor conflicts and
confusing ordering issues.
### Implementation
Runtime requirements have two additional challenges in comparison to the
`require` attribute.
1. The `require` attribute uses recursion and macros with clever
ordering to populate hash maps of required components for each component
type. The expected semantics are that "more specific" requirements
override ones deeper in the inheritance tree. However, at runtime, there
is no representation of how "specific" each requirement is.
2. If you first register the requirement `X -> Y`, and later register `Y
-> Z`, then `X` should also indirectly require `Z`. However, `Y` itself
doesn't know that it is required by `X`, so it's not aware that it
should update the list of required components for `X`.
My solutions to these problems are:
1. Store the depth in the inheritance tree for each entry of a given
component's `RequiredComponents`. This is used to determine how
"specific" each requirement is. For `require`-based registration, these
depths are computed as part of the recursion.
2. Store and maintain a `required_by` list in each component's
`ComponentInfo`, next to `required_components`. For `require`-based
registration, these are also added after each registration, as part of
the recursion.
When calling `register_required_components`, it works as follows:
1. Get the required components of `Foo`, and check that `Bar` isn't
already a *direct* requirement.
3. Register `Bar` as a required component for `Foo`, and add `Foo` to
the `required_by` list for `Bar`.
4. Find and register all indirect requirements inherited from `Bar`,
adding `Foo` to the `required_by` list for each component.
5. Iterate through components that require `Foo`, registering the new
inherited requires for them as indirect requirements.
The runtime registration is likely slightly more expensive than the
`require` version, but it is a one-time cost, and quite negligible in
practice, unless projects have hundreds or thousands of runtime
requirements. I have not benchmarked this however.
This does also add a small amount of extra cost to the `require`
attribute for updating `required_by` lists, but I expect it to be very
minor.
## Testing
I added some tests that are copies of the `require` versions, as well as
some tests that are more specific to the runtime implementation. I might
add a few more tests though.
## Discussion
- Is `register_required_components` a good name? Originally I went for
`register_component_requirement` to be consistent with
`register_component_hooks`, but the general feature is often referred to
as "required components", which is why I changed it to
`register_required_components`.
- Should we *not* panic for duplicate requirements? If so, should they
just be ignored, or should the latest registration overwrite earlier
ones?
- If we do want to panic for duplicate, conflicting registrations,
should we at least not panic if the registrations are *exactly* the
same, i.e. same component and same constructor? The current
implementation panics for all duplicate direct registrations regardless
of the constructor.
## Next Steps
- Allow `register_required_components` to take a `Bundle` instead of a
single required component.
- I could also try to do it in this PR if that would be preferable.
- Not directly related, but archetype invariants?
- Adopted from #14449
- Still fixes#12144.
## Migration Guide
The retained render world is a complex change: migrating might take one
of a few different forms depending on the patterns you're using.
For every example, we specify in which world the code is run. Most of
the changes affect render world code, so for the average Bevy user who's
using Bevy's high-level rendering APIs, these changes are unlikely to
affect your code.
### Spawning entities in the render world
Previously, if you spawned an entity with `world.spawn(...)`,
`commands.spawn(...)` or some other method in the rendering world, it
would be despawned at the end of each frame. In 0.15, this is no longer
the case and so your old code could leak entities. This can be mitigated
by either re-architecting your code to no longer continuously spawn
entities (like you're used to in the main world), or by adding the
`bevy_render::world_sync::TemporaryRenderEntity` component to the entity
you're spawning. Entities tagged with `TemporaryRenderEntity` will be
removed at the end of each frame (like before).
### Extract components with `ExtractComponentPlugin`
```
// main world
app.add_plugins(ExtractComponentPlugin::<ComponentToExtract>::default());
```
`ExtractComponentPlugin` has been changed to only work with synced
entities. Entities are automatically synced if `ComponentToExtract` is
added to them. However, entities are not "unsynced" if any given
`ComponentToExtract` is removed, because an entity may have multiple
components to extract. This would cause the other components to no
longer get extracted because the entity is not synced.
So be careful when only removing extracted components from entities in
the render world, because it might leave an entity behind in the render
world. The solution here is to avoid only removing extracted components
and instead despawn the entire entity.
### Manual extraction using `Extract<Query<(Entity, ...)>>`
```rust
// in render world, inspired by bevy_pbr/src/cluster/mod.rs
pub fn extract_clusters(
mut commands: Commands,
views: Extract<Query<(Entity, &Clusters, &Camera)>>,
) {
for (entity, clusters, camera) in &views {
// some code
commands.get_or_spawn(entity).insert(...);
}
}
```
One of the primary consequences of the retained rendering world is that
there's no longer a one-to-one mapping from entity IDs in the main world
to entity IDs in the render world. Unlike in Bevy 0.14, Entity 42 in the
main world doesn't necessarily map to entity 42 in the render world.
Previous code which called `get_or_spawn(main_world_entity)` in the
render world (`Extract<Query<(Entity, ...)>>` returns main world
entities). Instead, you should use `&RenderEntity` and
`render_entity.id()` to get the correct entity in the render world. Note
that this entity does need to be synced first in order to have a
`RenderEntity`.
When performing manual abstraction, this won't happen automatically
(like with `ExtractComponentPlugin`) so add a `SyncToRenderWorld` marker
component to the entities you want to extract.
This results in the following code:
```rust
// in render world, inspired by bevy_pbr/src/cluster/mod.rs
pub fn extract_clusters(
mut commands: Commands,
views: Extract<Query<(&RenderEntity, &Clusters, &Camera)>>,
) {
for (render_entity, clusters, camera) in &views {
// some code
commands.get_or_spawn(render_entity.id()).insert(...);
}
}
// in main world, when spawning
world.spawn(Clusters::default(), Camera::default(), SyncToRenderWorld)
```
### Looking up `Entity` ids in the render world
As previously stated, there's now no correspondence between main world
and render world `Entity` identifiers.
Querying for `Entity` in the render world will return the `Entity` id in
the render world: query for `MainEntity` (and use its `id()` method) to
get the corresponding entity in the main world.
This is also a good way to tell the difference between synced and
unsynced entities in the render world, because unsynced entities won't
have a `MainEntity` component.
---------
Co-authored-by: re0312 <re0312@outlook.com>
Co-authored-by: re0312 <45868716+re0312@users.noreply.github.com>
Co-authored-by: Periwink <charlesbour@gmail.com>
Co-authored-by: Anselmo Sampietro <ans.samp@gmail.com>
Co-authored-by: Emerson Coskey <56370779+ecoskey@users.noreply.github.com>
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
Co-authored-by: Christian Hughes <9044780+ItsDoot@users.noreply.github.com>
# Objective
Unlike `Capsule3d` which has the `.to_cylinder` method, `Capsule2d`
doesn't have an equivalent `.to_inner_rectangle` method and as shown by
#15191 this is surprisingly easy to get wrong
## Solution
Implemented a `Capsule2d::to_inner_rectangle` method as it is
implemented in the fixed `Capsule2d` shape sampling, and as I was adding
tests I noticed `Capsule2d` didn't implement `Measure2d` so I did this
as well.
## Changelog
### Added
- `Capsule2d::to_inner_rectangle`, `Capsule2d::area` and
`Capsule2d::perimeter`
---------
Co-authored-by: Joona Aalto <jondolf.dev@gmail.com>
Co-authored-by: James Liu <contact@jamessliu.com>
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
# Objective
`ui_stack_system` generates a tree of `StackingContexts` which it then
flattens to get the `UiStack`.
But there's no need to construct a new tree. We can query for nodes with
a global `ZIndex`, add those nodes to the root nodes list and then build
the `UiStack` from a walk of the existing layout tree, ignoring any
branches that have a global `Zindex`.
Fixes#9877
## Solution
Split the `ZIndex` enum into two separate components, `ZIndex` and
`GlobalZIndex`
Query for nodes with a `GlobalZIndex`, add those nodes to the root nodes
list and then build the `UiStack` from a walk of the existing layout
tree, filtering branches by `Without<GlobalZIndex>` so we don't revisit
nodes.
```
cargo run --profile stress-test --features trace_tracy --example many_buttons
```
<img width="672" alt="ui-stack-system-walk-split-enum"
src="https://github.com/bevyengine/bevy/assets/27962798/11e357a5-477f-4804-8ada-c4527c009421">
(Yellow is this PR, red is main)
---
## Changelog
`Zindex`
* The `ZIndex` enum has been split into two separate components `ZIndex`
(which replaces `ZIndex::Local`) and `GlobalZIndex` (which replaces
`ZIndex::Global`). An entity can have both a `ZIndex` and
`GlobalZIndex`, in comparisons `ZIndex` breaks ties if two
`GlobalZIndex` values are equal.
`ui_stack_system`
* Instead of generating a tree of `StackingContexts`, query for nodes
with a `GlobalZIndex`, add those nodes to the root nodes list and then
build the `UiStack` from a walk of the existing layout tree, filtering
branches by `Without<GlobalZIndex` so we don't revisit nodes.
## Migration Guide
The `ZIndex` enum has been split into two separate components `ZIndex`
(which replaces `ZIndex::Local`) and `GlobalZIndex` (which replaces
`ZIndex::Global`). An entity can have both a `ZIndex` and
`GlobalZIndex`, in comparisons `ZIndex` breaks ties if two
`GlobalZindex` values are equal.
---------
Co-authored-by: Gabriel Bourgeois <gabriel.bourgeoisv4si@gmail.com>
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
Co-authored-by: UkoeHB <37489173+UkoeHB@users.noreply.github.com>
(Note: #15434 implements something very similar to this for functional
curve adaptors, which is why they aren't present in this PR.)
# Objective
Previously, there was basically no chance that the
explicitly-interpolating sample curve structs from the `Curve` API would
actually be `Reflect`. The reason for this is functional programming:
the structs contain an explicit interpolation `I: Fn(&T, &T, f32) -> T`
which, under typical circumstances, will never be `Reflect`, which
prevents the derive from realistically succeeding. In fact, they won't
be a lot of other things either, notably including both`Debug` and
`TypePath`, which are also required for reflection to succeed.
The goal of this PR is to weaken the implementations of reflection
traits for these structs so that they can implement `Reflect` under
reasonable circumstances. (Notably, they will still not be
`FromReflect`, which is unavoidable.)
## Solution
The function fields are marked as `#[reflect(ignore)]`, and the derive
macro for `Reflect` has `FromReflect` disabled. (This is not fully
optimal, but we don't presently have any kind of "read-only" attribute
for these fields.) Additionally, these structs receive custom `Debug`
and `TypePath` implementations that display the function's (unstable!)
type name instead of its value or type path (respectively). In the case
of `TypePath`, this is a bit janky, but the instability of `type_name`
won't generally present an issue for generics, which would have to be
registered manually in the type registry anyway, which is impossible
because the function type parameters cannot be named.
(And in general, the "blessed" route for such cases would generally
involve manually monomorphizing the function parameter away, which also
allows access to `FromReflect` etc. through very ordinary use of the
derive macro.)
## Testing
Tests in the new `bevy_math::curve::sample_curves` module guarantee that
these are actually `Reflect` under reasonable circumstances.
---
## Future changes
If and when function item types become `Default`, these types will need
to receive custom `FromReflect` implementations that exploit it. Such a
custom implementation would also be desirable if users start doing
things like wrapping function items in `Default`/`FromReflect` wrappers
that still implement a `Fn` trait.
Additionally, if function types become nameable in user-space, the
stance on `Debug`/`TypePath` may bear reexamination, since partial
monomorphization through wrappers would make implementing reflect traits
for function types potentially more viable.
---------
Co-authored-by: Gino Valente <49806985+MrGVSV@users.noreply.github.com>
# Objective
Fix#12273
## Solution
– Only emit `KeyboardFocusLost` when the keyboard focus is lost
– ignore synthetic key releases too, not just key presses (as they're
already covered by `KeyboardFocusLost`)
---
## Changelog
### Fixed
- Don't trigger `ButtonInput<KeyCode>::just_pressed`/`just_released`
when spawning a window/focus moving between Bevy windows
# Objective
- Contributes to #15460
- Allows `bevy_mikktspace` to be used in `no_std` contexts.
## Solution
- Added `std` (default) and `libm` features which control the inclusion
of the standard library. To use `bevy_mikktspace` in `no_std`
environments, enable the `libm` feature.
## Testing
- CI
- `cargo clippy -p bevy_mikktspace --target "x86_64-unknown-none"
--no-default-features --features libm`
# Objective
Add a `Populated` system parameter that acts like `Query`, but prevents
system from running if there are no matching entities.
Fixes: #15302
## Solution
Implement the system param which newtypes the `Query`.
The only change is new validation, which fails if query is empty.
The new system param is used in `fallible_params` example.
## Testing
Ran `fallible_params` example.
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
# Objective
Currently, reflecting a generic type provides no information about the
generic parameters. This means that you can't get access to the type of
`T` in `Foo<T>` without creating custom type data (we do this for
[`ReflectHandle`](https://docs.rs/bevy/0.14.2/bevy/asset/struct.ReflectHandle.html#method.asset_type_id)).
## Solution
This PR makes it so that generic type parameters and generic const
parameters are tracked in a `Generics` struct stored on the `TypeInfo`
for a type.
For example, `struct Foo<T, const N: usize>` will store `T` and `N` as a
`TypeParamInfo` and `ConstParamInfo`, respectively.
The stored information includes:
- The name of the generic parameter (i.e. `T`, `N`, etc.)
- The type of the generic parameter (remember that we're dealing with
monomorphized types, so this will actually be a concrete type)
- The default type/value, if any (e.g. `f32` in `T = f32` or `10` in
`const N: usize = 10`)
### Caveats
The only requirement for this to work is that the user does not opt-out
of the automatic `TypePath` derive with `#[reflect(type_path = false)]`.
Doing so prevents the macro code from 100% knowing that the generic type
implements `TypePath`. This in turn means the generated `Typed` impl
can't add generics to the type.
There are two solutions for this—both of which I think we should explore
in a future PR:
1. We could just not use `TypePath`. This would mean that we can't store
the `Type` of the generic, but we can at least store the `TypeId`.
2. We could provide a way to opt out of the automatic `Typed` derive
with a `#[reflect(typed = false)]` attribute. This would allow users to
manually implement `Typed` to add whatever generic information they need
(e.g. skipping a parameter that can't implement `TypePath` while the
rest can).
I originally thought about making `Generics` an enum with `Generic`,
`NonGeneric`, and `Unavailable` variants to signify whether there are
generics, no generics, or generics that cannot be added due to opting
out of `TypePath`. I ultimately decided against this as I think it adds
a bit too much complexity for such an uncommon problem.
Additionally, user's don't necessarily _have_ to know the generics of a
type, so just skipping them should generally be fine for now.
## Testing
You can test locally by running:
```
cargo test --package bevy_reflect
```
---
## Showcase
You can now access generic parameters via `TypeInfo`!
```rust
#[derive(Reflect)]
struct MyStruct<T, const N: usize>([T; N]);
let generics = MyStruct::<f32, 10>::type_info().generics();
// Get by index:
let t = generics.get(0).unwrap();
assert_eq!(t.name(), "T");
assert!(t.ty().is::<f32>());
assert!(!t.is_const());
// Or by name:
let n = generics.get_named("N").unwrap();
assert_eq!(n.name(), "N");
assert!(n.ty().is::<usize>());
assert!(n.is_const());
```
You can even access parameter defaults:
```rust
#[derive(Reflect)]
struct MyStruct<T = String, const N: usize = 10>([T; N]);
let generics = MyStruct::<f32, 5>::type_info().generics();
let GenericInfo::Type(info) = generics.get_named("T").unwrap() else {
panic!("expected a type parameter");
};
let default = info.default().unwrap();
assert!(default.is::<String>());
let GenericInfo::Const(info) = generics.get_named("N").unwrap() else {
panic!("expected a const parameter");
};
let default = info.default().unwrap();
assert_eq!(default.downcast_ref::<usize>().unwrap(), &10);
```
# Objective
We introduced the fancy Curve API earlier in this version. The goal of
this PR is to provide a level of integration between that API and the
existing spline constructions in `bevy_math`.
Note that this PR only covers the integration of position-sampling via
the `Curve` API. Other (substantially more complex) planned work will
introduce general facilities for handling derivatives.
## Solution
`CubicSegment`, `CubicCurve`, `RationalSegment`, and `RationalCurve` all
now implement `Curve`, using their `position` function to sample the
output.
Additionally, some documentation has been updated/corrected, and
`Serialize`/`Deserialize` derives have been added for all the curve
structs. (Note that there are some barriers to automatic registration of
`ReflectSerialize`/`ReflectSerialize` involving generics that have not
been resolved in this PR.)
---
## Migration Guide
The `RationalCurve::domain` method has been renamed to
`RationalCurve::length`. Calling `.domain()` on a `RationalCurve` now
returns its entire domain as an `Interval`.
# Objective
- The shader_instancing example can be misleading since it doesn't
explain that bevy has built in automatic instancing.
## Solution
- Explain that bevy has built in instancing and that this example is for
advanced users.
- Add a new automatic_instancing example that shows how to use the built
in automatic instancing
- Rename the shader_instancing example to custom_shader_instancing to
highlight that this is a more advanced implementation
---------
Co-authored-by: JMS55 <47158642+JMS55@users.noreply.github.com>
# Objective
* Remove all uses of render_resource_wrapper.
* Make it easier to share a `wgpu::Device` between Bevy and application
code.
## Solution
Removed the `render_resource_wrapper` macro.
To improve the `RenderCreation:: Manual ` API, `ErasedRenderDevice` was
replaced by `Arc`. Unfortunately I had to introduce one more usage of
`WgpuWrapper` which seems like an unwanted constraint on the caller.
## Testing
- Did you test these changes? If so, how?
- Ran `cargo test`.
- Ran a few examples.
- Used `RenderCreation::Manual` in my own project
- Exercised `RenderCreation::Automatic` through examples
- Are there any parts that need more testing?
- No
- How can other people (reviewers) test your changes? Is there anything
specific they need to know?
- Run examples
- Use `RenderCreation::Manual` in their own project
# Objective
- Provide a generic and _reflectable_ way to iterate over contained
entities
## Solution
Adds two new traits:
* `VisitEntities`: Reflectable iteration, accepts a closure rather than
producing an iterator. Implemented by default for `IntoIterator`
implementing types. A proc macro is also provided.
* A `Mut` variant of the above. Its derive macro uses the same field
attribute to avoid repetition.
## Testing
Added a test for `VisitEntities` that also transitively tests its derive
macro as well as the default `MapEntities` impl.
# Objective
Adds a new `Readback` component to request for readback of a
`Handle<Image>` or `Handle<ShaderStorageBuffer>` to the CPU in a future
frame.
## Solution
We track the `Readback` component and allocate a target buffer to write
the gpu resource into and map it back asynchronously, which then fires a
trigger on the entity in the main world. This proccess is asynchronous,
and generally takes a few frames.
## Showcase
```rust
let mut buffer = ShaderStorageBuffer::from(vec![0u32; 16]);
buffer.buffer_description.usage |= BufferUsages::COPY_SRC;
let buffer = buffers.add(buffer);
commands
.spawn(Readback::buffer(buffer.clone()))
.observe(|trigger: Trigger<ReadbackComplete>| {
info!("Buffer data from previous frame {:?}", trigger.event());
});
```
---------
Co-authored-by: Kristoffer Søholm <k.soeholm@gmail.com>
Co-authored-by: IceSentry <IceSentry@users.noreply.github.com>
# Objective
`World::flush_commands` will cause a panic with `error[B0003]: Could not
insert a bundle [...] for entity [...] because it doesn't exist in this
World` if there was a `spawn` command in the queue and you should
instead use `flush` for this but this isn't mentioned in the docs
## Solution
Add a note to the docs suggesting to use `World::flush` in this context.
This error doesn't appear to happen with `spawn_batch` so I didn't add
that to the note although you can cause it with
`commands.spawn_empty().insert(...)` but I wasn't sure that was worth
the documentation complexity as it is pretty unlikely (and equivalent to
`commands.spawn(...)`.
# Objective
Fixes#15185.
# Solution
Change `drain` to take a `&mut self` for most reflected types.
Some notable exceptions to this change are `Array` and `Tuple`. These
types don't make sense with `drain` taking a mutable borrow since they
can't get "smaller". Also `BTreeMap` doesn't have a `drain` function, so
we have to pop elements off one at a time.
## Testing
- The existing tests are sufficient.
---
## Migration Guide
- `reflect::Map`, `reflect::List`, and `reflect::Set` all now take a
`&mut self` instead of a `Box<Self>`. Callers of these traits should add
`&mut` before their boxes, and implementers of these traits should
update to match.
# Objective
Mostly covers the first point in
https://github.com/bevyengine/bevy/issues/13713#issuecomment-2364786694
The idea here is that a lot of people want to load their own texture
atlases, and many of them do this by deserializing some custom version
of `TextureAtlasLayout`. This makes that a little easier by providing
`serde` impls for them.
## Solution
In order to make `TextureAtlasLayout` serializable, the custom texture
mappings that are added by `TextureAtlasBuilder` were separated into
their own type, `TextureAtlasSources`. The inner fields are made public
so people can create their own version of this type, although because it
embeds asset IDs, it's not as easily serializable. In particular,
atlases that are loaded directly (e.g. sprite sheets) will not have a
copy of this map, and so, don't need to construct it at all.
As an aside, since this is the very first thing in `bevy_sprite` with
`serde` impls, I've added a `serialize` feature to the crate and made
sure it gets activated when the `serialize` feature is enabled on the
parent `bevy` crate.
## Testing
I was kind of shocked that there isn't anywhere in the code besides a
single example that actually used this functionality, so, it was
relatively straightforward to do.
In #13713, among other places, folks have mentioned adding custom
serialization into their pipelines. It would be nice to hear from people
whether this change matches what they're doing in their code, and if
it's relatively seamless to adapt to. I suspect that the answer is yes,
but, that's mainly the only other kind of testing that can be added.
## Migration Guide
`TextureAtlasBuilder` no longer stores a mapping back to the original
images in `TextureAtlasLayout`; that functionality has been added to a
new struct, `TextureAtlasSources`, instead. This also means that the
signature for `TextureAtlasBuilder::finish` has changed, meaning that
calls of the form:
```rust
let (atlas_layout, image) = builder.build()?;
```
Will now change to the form:
```rust
let (atlas_layout, atlas_sources, image) = builder.build()?;
```
And instead of performing a reverse-lookup from the layout, like so:
```rust
let atlas_layout_handle = texture_atlases.add(atlas_layout.clone());
let index = atlas_layout.get_texture_index(&my_handle);
let handle = TextureAtlas {
layout: atlas_layout_handle,
index,
};
```
You can perform the lookup from the sources instead:
```rust
let atlas_layout = texture_atlases.add(atlas_layout);
let index = atlas_sources.get_texture_index(&my_handle);
let handle = TextureAtlas {
layout: atlas_layout,
index,
};
```
Additionally, `TextureAtlasSources` also has a convenience method,
`handle`, which directly combines the index and an existing
`TextureAtlasLayout` handle into a new `TextureAtlas`:
```rust
let atlas_layout = texture_atlases.add(atlas_layout);
let handle = atlas_sources.handle(atlas_layout, &my_handle);
```
## Extra notes
In the future, it might make sense to combine the three types returned
by `TextureAtlasBuilder` into their own struct, just so that people
don't need to assign variable names to all three parts. In particular,
when creating a version that can be loaded directly (like #11873), we
could probably use this new type.
# Objective
- Fix panic when atlas index is out of bounds
- Took the chance to clean it up a bit
## Solution
- Use texture dimensions like rendering pipeline. Dropped atlas layouts
and indexes out of bounds are shown as a sprite.
## Testing
Used sprite_picking example, drop layout and/or use indexes out of
bounds.
# Objective
- `DynamicMap` currently uses an `HashMap` from a `u64` hash to the
entry index in a `Vec`. This is incorrect in the presence of hash
collisions, so let's fix it;
- `DynamicMap::remove` was also buggy, as it didn't fix up the indexes
of the other elements after removal. Fix that up as well and add a
regression test.
## Solution
- Use `HashTable` in `DynamicMap` to distinguish entries that have the
same hash by using `reflect_partial_eq`, bringing it more in line with
what `DynamicSet` does;
- Reimplement `DynamicMap::remove` to properly fix up the index of moved
elements after the removal.
## Testing
- A regression test was added for the `DynamicMap::remove` issue.
---
Some kinda related considerations: the use of a separate `Vec` for
storing the entries adds some complications that I'm not sure are worth.
This is mainly used to implement an efficient `get_at`, which is relied
upon by `MapIter`. However both `HashMap` and `BTreeMap` implement
`get_at` inefficiently (and cannot do so efficiently), leading to a
`O(N^2)` complexity for iterating them. This could be removed in favor
of a `Box<dyn Iterator>` like it's done in `DynamicSet`.
# Objective
Improve the documentation of `SystemParamBuilder`. Not all builder types
have documentation, and the documentation is spread around and not
linked together well.
## Solution
Reorganize `SystemParamBuilder` docs and examples. All builder types now
have their own examples, and the list of builder types is linked from
the `SystemParamBuilder` trait. Add some examples to `FilteredEntityRef`
and `FilteredEntityMut` so that `QueryParamBuilder` can reference them.
# Objective
- Fixes#15077
## Solution
- Clears `ViewTargetAttachments` resource every frame before
`create_surfaces` system instead, which was previously done after
`extract_windows`.
## Testing
- Confirmed that examples no longer panic on window resizing with DX12
backend.
- `screenshot` example keeps working after this change.
# Objective
This implements another item on the way to complete the `Curves`
implementation initiative
Citing @mweatherley
> Curve adaptors for making a curve repeat or ping-pong would be useful.
This adds three widely applicable adaptors:
- `ReverseCurve` "plays" the curve backwards
- `RepeatCurve` to repeat the curve for `n` times where `n` in `[0,inf)`
- `ForeverCurve` which extends the curves domain to `EVERYWHERE`
- `PingPongCurve` (name wip (?)) to chain the curve with it's reverse.
This would be achievable with `ReverseCurve` and `ChainCurve`, but it
would require the use of `by_ref` which can be restrictive in some
scenarios where you'd rather just consume the curve. Users can still
create the same effect by combination of the former two, but since this
will be most likely a very typical adaptor we should also provide it on
the library level. (Why it's typical: you can create a single period of
common waves with it pretty easily, think square wave (= pingpong +
step), triangle wave ( = pingpong + linear), etc.)
- `ContinuationCurve` which chains two curves but also makes sure that
the samples of the second curve are translated so that
`sample(first.end) == sample(second.start)`
## Solution
Implement the adaptors above. (More suggestions are welcome!)
## Testing
- [x] add simple tests. One per adaptor
---------
Co-authored-by: eckz <567737+eckz@users.noreply.github.com>
Co-authored-by: Matty <2975848+mweatherley@users.noreply.github.com>
Co-authored-by: IQuick 143 <IQuick143cz@gmail.com>
Co-authored-by: Matty <weatherleymatthew@gmail.com>
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
# Objective
- fixes https://github.com/bevyengine/bevy/issues/13473
## Solution
- When a single mesh is assigned multiple materials, it is divided into
several primitive nodes, with each primitive assigned a unique material.
Presently, these primitives are named using the format Mesh.index, which
complicates querying. To improve this, we can assign a specific name to
each primitive based on the material’s name, since each primitive
corresponds to one material exclusively.
## Testing
- I have included a simple example which shows how to query a material
and mesh part based on the new name component.
## Changelog
- adds `GltfMaterialName` component to the mesh entity of the gltf
primitive node.
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
# Objective
Fixes#15394
## Solution
Observers now validate params.
System registry has a new error variant for when system running fails
due to invalid parameters.
Run once now returns a `Result<Out, RunOnceError>` instead of `Out`.
This is more inline with system registry, which also returns a result.
I'll address warning messages in #15500.
## Testing
Added one test for each case.
---
## Migration Guide
- `RunSystemOnce::run_system_once` and
`RunSystemOnce::run_system_once_with` now return a `Result<Out>` instead
of just `Out`
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
Co-authored-by: Zachary Harrold <zac@harrold.com.au>
* Save 16 bytes per vertex by calculating tangents in the shader at
runtime, rather than storing them in the vertex data.
* Based on https://jcgt.org/published/0009/03/04,
https://www.jeremyong.com/graphics/2023/12/16/surface-gradient-bump-mapping.
* Fixed visbuffer resolve to use the updated algorithm that flips ddy
correctly
* Added some more docs about meshlet material limitations, and some
TODOs about transforming UV coordinates for the future.
For testing add a normal map to the bunnies with StandardMaterial like
below, and then test that on both main and this PR (make sure to
download the correct bunny for each). Results should be mostly
identical.
```rust
normal_map_texture: Some(asset_server.load_with_settings(
"textures/BlueNoise-Normal.png",
|settings: &mut ImageLoaderSettings| settings.is_srgb = false,
)),
```
# Objective
- Resolves#15453
## Solution
- Added new `World::resource_id` and `World::register_resource` methods
to support this feature
- Added new `ReflectResource::register_resource` method, and new pointer
to this new function
- Added new `ReflectComponent::register_component`
## Testing
- Tested this locally, but couldn't test the entire crate locally, just
this new feature, expect that CI will do the rest of the work.
---
## Showcase
```rs
#[derive(Component, Reflect)]
#[reflect(Component)]
struct MyComp;
let mut world = World::new();
let mut registry = TypeRegistration::of::<MyComp>();
registry.insert::<ReflectComponent>(FromType::<MyComp>::from_type());
let data = registry.data::<ReflectComponent>().unwrap();
// Its now possible to register the Component in the world this way
let component_id = data.register_component(&mut world);
// They will be the same
assert_eq!(component_id, world.component_id::<MyComp>().unwrap());
```
```rs
#[derive(Resource, Reflect)]
#[reflect(Resource)]
struct MyResource;
let mut world = World::new();
let mut registry = TypeRegistration::of::<MyResource>();
registry.insert::<ReflectResource>(FromType::<MyResource>::from_type());
let data = registry.data::<ReflectResource>().unwrap();
// Same with resources
let component_id = data.register_resource(&mut world);
// They match
assert_eq!(component_id, world.resource_id::<MyResource>().unwrap());
```
# Objective
Add the following system params:
- `QuerySingle<D, F>` - Valid if only one matching entity exists,
- `Option<QuerySingle<D, F>>` - Valid if zero or one matching entity
exists.
As @chescock pointed out, we don't need `Mut` variants.
Fixes: #15264
## Solution
Implement the type and both variants of system params.
Also implement `ReadOnlySystemParam` for readonly queries.
Added a new ECS example `fallible_params` which showcases `SingleQuery`
usage.
In the future we might want to add `NonEmptyQuery`,
`NonEmptyEventReader` and `Res` to it (or maybe just stop at mentioning
it).
## Testing
Tested with the example.
There is a lot of warning spam so we might want to implement #15391.
# Objective
- We use a feature introduced in async-channel 2.3.0, `force_send`
- Existing project fail to compile as they have a lock file on the 2.2.X
## Solution
- Bump async-channel
> [!NOTE]
> This is my first PR, so if something is incorrect
> or missing, please let me know :3
# Objective
- Clarifies `spawn`, `spawn_batch` and `ParallelCommands` docs about
performance and use cases
- Fixes#15472
## Solution
Add comments to `spawn`, `spawn_batch` and `ParallelCommands` to clarify
the
intended use case and link to other/better ways of doing spawning things
for
certain use cases.
# Objective
- Fixes#15447
## Solution
- Remove the `Return::Unit` variant and use a `Return::Owned` variant
holding a unit `()` type.
## Migration Guide
- Removed the `Return::Unit` variant; use `Return::unit()` instead.
---------
Co-authored-by: Gino Valente <49806985+MrGVSV@users.noreply.github.com>
# Objective
The "zero-sized" description was outdated and misleading.
## Solution
Changed the description to just say that it's an enumeration (an enum)
# Objective
- Fixes#15490 introduced in #15094.
## Solution
- Use non-panicking `try_insert`
## Testing
- Closing window with `CursorIcon` no longer crashes after this change
(confirmed with `window_settings` example)
# Objective
- Contributes to #15460
## Solution
- Wrap `derive_label` `quote!` in an anonymous constant which contains
an `extern crate alloc` statement, allowing use of the `alloc` namespace
even when a user has not brought in the crate themselves.
## Testing
- CI passed locally.
## Notes
We can't generate code that uses `::std::boxed::Box` in `no_std`
environments, but we also can't rely on `::alloc::boxed::Box` either,
since the user might not have declared `extern crate alloc`. To resolve
this, the generated code is wrapped in an anonymous constant which
contains the `extern crate alloc` invocation.
This does mean the macro is no longer hygienic against cases where the
user provides an alternate `alloc` crate, however I believe this is an
acceptable compromise.
Additionally, this crate itself doesn't need to be `no_std`, it just
needs to _generate_ `no_std` compatible code.
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
## Objective
Closes#15408 (somewhat)
## Solution
- Moved the existing HTTP transport to its own module with its own
plugin (`RemoteHttpPlugin`) (disabled on WASM)
- Swapped out the `smol` crate for the smaller crates it re-exports to
make it easier to keep out non-wasm code (HTTP transport needs
`async-io` which can't build on WASM)
- Added a new public `BrpSender` resource holding the matching sender
for the `BrpReceiver`' (formally `BrpMailbox`). This allows other crates
to send `BrpMessage`'s to the "mailbox".
## Testing
TODO
---------
Co-authored-by: Matty <weatherleymatthew@gmail.com>
# Objective
- Significantly improve the ergonomics of gamepads and allow new
features
Gamepads are a bit unergonomic to work with, they use resources but
unlike other inputs, they are not limited to a single gamepad, to get
around this it uses an identifier (Gamepad) to interact with anything
causing all sorts of issues.
1. There are too many: Gamepads, GamepadSettings, GamepadInfo,
ButtonInput<T>, 2 Axis<T>.
2. ButtonInput/Axis generic methods become really inconvenient to use
e.g. any_pressed()
3. GamepadButton/Axis structs are unnecessary boilerplate:
```rust
for gamepad in gamepads.iter() {
if button_inputs.just_pressed(GamepadButton::new(gamepad, GamepadButtonType::South)) {
info!("{:?} just pressed South", gamepad);
} else if button_inputs.just_released(GamepadButton::new(gamepad, GamepadButtonType::South))
{
info!("{:?} just released South", gamepad);
}
}
```
4. Projects often need to create resources to store the selected gamepad
and have to manually check if their gamepad is still valid anyways.
- Previously attempted by #3419 and #12674
## Solution
- Implement gamepads as entities.
Using entities solves all the problems above and opens new
possibilities.
1. Reduce boilerplate and allows iteration
```rust
let is_pressed = gamepads_buttons.iter().any(|buttons| buttons.pressed(GamepadButtonType::South))
```
2. ButtonInput/Axis generic methods become ergonomic again
```rust
gamepad_buttons.any_just_pressed([GamepadButtonType::Start, GamepadButtonType::Select])
```
3. Reduces the number of public components significantly (Gamepad,
GamepadSettings, GamepadButtons, GamepadAxes)
4. Components are highly convenient. Gamepad optional features could now
be expressed naturally (`Option<Rumble> or Option<Gyro>`), allows devs
to attach their own components and filter them, so code like this
becomes possible:
```rust
fn move_player<const T: usize>(
player: Query<&Transform, With<Player<T>>>,
gamepads_buttons: Query<&GamepadButtons, With<Player<T>>>,
) {
if let Ok(gamepad_buttons) = gamepads_buttons.get_single() {
if gamepad_buttons.pressed(GamepadButtonType::South) {
// move player
}
}
}
```
---
## Follow-up
- [ ] Run conditions?
- [ ] Rumble component
# Changelog
## Added
TODO
## Changed
TODO
## Removed
TODO
## Migration Guide
TODO
---------
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
The next step in the migration to required components: Deprecate
`VisibilityBundle` and make `Visibility` require `InheritedVisibility`
and `ViewVisibility`, as per the [chosen
proposal](https://hackmd.io/@bevy/required_components/%2FcO7JPSAQR5G0J_j5wNwtOQ).
## Solution
Deprecate `VisibilityBundle` and make `Visibility` require
`InheritedVisibility` and `ViewVisibility`.
I chose not to deprecate `SpatialBundle` yet, as doing so would mean
that we need to manually add `Visibility` to a bunch of places. It will
be nicer once meshes, sprites, lights, fog, and cameras have been
migrated, since they will require `Transform` and `Visibility` and
therefore not need manually added defaults for them.
---
## Migration Guide
Replace all insertions of `VisibilityBundle` with the `Visibility`
component. The other components required by it will now be inserted
automatically.
