This adds "high level camera driven rendering" to Bevy. The goal is to give users more control over what gets rendered (and where) without needing to deal with render logic. This will make scenarios like "render to texture", "multiple windows", "split screen", "2d on 3d", "3d on 2d", "pass layering", and more significantly easier.
Here is an [example of a 2d render sandwiched between two 3d renders (each from a different perspective)](https://gist.github.com/cart/4fe56874b2e53bc5594a182fc76f4915):
![image](https://user-images.githubusercontent.com/2694663/168411086-af13dec8-0093-4a84-bdd4-d4362d850ffa.png)
Users can now spawn a camera, point it at a RenderTarget (a texture or a window), and it will "just work".
Rendering to a second window is as simple as spawning a second camera and assigning it to a specific window id:
```rust
// main camera (main window)
commands.spawn_bundle(Camera2dBundle::default());
// second camera (other window)
commands.spawn_bundle(Camera2dBundle {
camera: Camera {
target: RenderTarget::Window(window_id),
..default()
},
..default()
});
```
Rendering to a texture is as simple as pointing the camera at a texture:
```rust
commands.spawn_bundle(Camera2dBundle {
camera: Camera {
target: RenderTarget::Texture(image_handle),
..default()
},
..default()
});
```
Cameras now have a "render priority", which controls the order they are drawn in. If you want to use a camera's output texture as a texture in the main pass, just set the priority to a number lower than the main pass camera (which defaults to `0`).
```rust
// main pass camera with a default priority of 0
commands.spawn_bundle(Camera2dBundle::default());
commands.spawn_bundle(Camera2dBundle {
camera: Camera {
target: RenderTarget::Texture(image_handle.clone()),
priority: -1,
..default()
},
..default()
});
commands.spawn_bundle(SpriteBundle {
texture: image_handle,
..default()
})
```
Priority can also be used to layer to cameras on top of each other for the same RenderTarget. This is what "2d on top of 3d" looks like in the new system:
```rust
commands.spawn_bundle(Camera3dBundle::default());
commands.spawn_bundle(Camera2dBundle {
camera: Camera {
// this will render 2d entities "on top" of the default 3d camera's render
priority: 1,
..default()
},
..default()
});
```
There is no longer the concept of a global "active camera". Resources like `ActiveCamera<Camera2d>` and `ActiveCamera<Camera3d>` have been replaced with the camera-specific `Camera::is_active` field. This does put the onus on users to manage which cameras should be active.
Cameras are now assigned a single render graph as an "entry point", which is configured on each camera entity using the new `CameraRenderGraph` component. The old `PerspectiveCameraBundle` and `OrthographicCameraBundle` (generic on camera marker components like Camera2d and Camera3d) have been replaced by `Camera3dBundle` and `Camera2dBundle`, which set 3d and 2d default values for the `CameraRenderGraph` and projections.
```rust
// old 3d perspective camera
commands.spawn_bundle(PerspectiveCameraBundle::default())
// new 3d perspective camera
commands.spawn_bundle(Camera3dBundle::default())
```
```rust
// old 2d orthographic camera
commands.spawn_bundle(OrthographicCameraBundle::new_2d())
// new 2d orthographic camera
commands.spawn_bundle(Camera2dBundle::default())
```
```rust
// old 3d orthographic camera
commands.spawn_bundle(OrthographicCameraBundle::new_3d())
// new 3d orthographic camera
commands.spawn_bundle(Camera3dBundle {
projection: OrthographicProjection {
scale: 3.0,
scaling_mode: ScalingMode::FixedVertical,
..default()
}.into(),
..default()
})
```
Note that `Camera3dBundle` now uses a new `Projection` enum instead of hard coding the projection into the type. There are a number of motivators for this change: the render graph is now a part of the bundle, the way "generic bundles" work in the rust type system prevents nice `..default()` syntax, and changing projections at runtime is much easier with an enum (ex for editor scenarios). I'm open to discussing this choice, but I'm relatively certain we will all come to the same conclusion here. Camera2dBundle and Camera3dBundle are much clearer than being generic on marker components / using non-default constructors.
If you want to run a custom render graph on a camera, just set the `CameraRenderGraph` component:
```rust
commands.spawn_bundle(Camera3dBundle {
camera_render_graph: CameraRenderGraph::new(some_render_graph_name),
..default()
})
```
Just note that if the graph requires data from specific components to work (such as `Camera3d` config, which is provided in the `Camera3dBundle`), make sure the relevant components have been added.
Speaking of using components to configure graphs / passes, there are a number of new configuration options:
```rust
commands.spawn_bundle(Camera3dBundle {
camera_3d: Camera3d {
// overrides the default global clear color
clear_color: ClearColorConfig::Custom(Color::RED),
..default()
},
..default()
})
commands.spawn_bundle(Camera3dBundle {
camera_3d: Camera3d {
// disables clearing
clear_color: ClearColorConfig::None,
..default()
},
..default()
})
```
Expect to see more of the "graph configuration Components on Cameras" pattern in the future.
By popular demand, UI no longer requires a dedicated camera. `UiCameraBundle` has been removed. `Camera2dBundle` and `Camera3dBundle` now both default to rendering UI as part of their own render graphs. To disable UI rendering for a camera, disable it using the CameraUi component:
```rust
commands
.spawn_bundle(Camera3dBundle::default())
.insert(CameraUi {
is_enabled: false,
..default()
})
```
## Other Changes
* The separate clear pass has been removed. We should revisit this for things like sky rendering, but I think this PR should "keep it simple" until we're ready to properly support that (for code complexity and performance reasons). We can come up with the right design for a modular clear pass in a followup pr.
* I reorganized bevy_core_pipeline into Core2dPlugin and Core3dPlugin (and core_2d / core_3d modules). Everything is pretty much the same as before, just logically separate. I've moved relevant types (like Camera2d, Camera3d, Camera3dBundle, Camera2dBundle) into their relevant modules, which is what motivated this reorganization.
* I adapted the `scene_viewer` example (which relied on the ActiveCameras behavior) to the new system. I also refactored bits and pieces to be a bit simpler.
* All of the examples have been ported to the new camera approach. `render_to_texture` and `multiple_windows` are now _much_ simpler. I removed `two_passes` because it is less relevant with the new approach. If someone wants to add a new "layered custom pass with CameraRenderGraph" example, that might fill a similar niche. But I don't feel much pressure to add that in this pr.
* Cameras now have `target_logical_size` and `target_physical_size` fields, which makes finding the size of a camera's render target _much_ simpler. As a result, the `Assets<Image>` and `Windows` parameters were removed from `Camera::world_to_screen`, making that operation much more ergonomic.
* Render order ambiguities between cameras with the same target and the same priority now produce a warning. This accomplishes two goals:
1. Now that there is no "global" active camera, by default spawning two cameras will result in two renders (one covering the other). This would be a silent performance killer that would be hard to detect after the fact. By detecting ambiguities, we can provide a helpful warning when this occurs.
2. Render order ambiguities could result in unexpected / unpredictable render results. Resolving them makes sense.
## Follow Up Work
* Per-Camera viewports, which will make it possible to render to a smaller area inside of a RenderTarget (great for something like splitscreen)
* Camera-specific MSAA config (should use the same "overriding" pattern used for ClearColor)
* Graph Based Camera Ordering: priorities are simple, but they make complicated ordering constraints harder to express. We should consider adopting a "graph based" camera ordering model with "before" and "after" relationships to other cameras (or build it "on top" of the priority system).
* Consider allowing graphs to run subgraphs from any nest level (aka a global namespace for graphs). Right now the 2d and 3d graphs each need their own UI subgraph, which feels "fine" in the short term. But being able to share subgraphs between other subgraphs seems valuable.
* Consider splitting `bevy_core_pipeline` into `bevy_core_2d` and `bevy_core_3d` packages. Theres a shared "clear color" dependency here, which would need a new home.
# Objective
- Bevy currently panics when displaying text with a *very* big font size (with font size greater than 400, the glyph would have a width or height greater than 512)
```
thread 'main' panicked at 'Fatal error when processing text: failed to add glyph to newly-created atlas GlyphId(514).', crates/bevy_ui/src/widget/text.rs:118:21
```
## Solution
- Create font atlas that scales up with the size of the glyphs
# Objective
- Split PBR and 2D mesh shaders into types and bindings to prepare the shaders to be more reusable.
- See #3969 for details. I'm doing this in multiple steps to make review easier.
---
## Changelog
- Changed: 2D and PBR mesh shaders are now split into types and bindings, the following shader imports are available: `bevy_pbr::mesh_view_types`, `bevy_pbr::mesh_view_bindings`, `bevy_pbr::mesh_types`, `bevy_pbr::mesh_bindings`, `bevy_sprite::mesh2d_view_types`, `bevy_sprite::mesh2d_view_bindings`, `bevy_sprite::mesh2d_types`, `bevy_sprite::mesh2d_bindings`
## Migration Guide
- In shaders for 3D meshes:
- `#import bevy_pbr::mesh_view_bind_group` -> `#import bevy_pbr::mesh_view_bindings`
- `#import bevy_pbr::mesh_struct` -> `#import bevy_pbr::mesh_types`
- NOTE: If you are using the mesh bind group at bind group index 2, you can remove those binding statements in your shader and just use `#import bevy_pbr::mesh_bindings` which itself imports the mesh types needed for the bindings.
- In shaders for 2D meshes:
- `#import bevy_sprite::mesh2d_view_bind_group` -> `#import bevy_sprite::mesh2d_view_bindings`
- `#import bevy_sprite::mesh2d_struct` -> `#import bevy_sprite::mesh2d_types`
- NOTE: If you are using the mesh2d bind group at bind group index 2, you can remove those binding statements in your shader and just use `#import bevy_sprite::mesh2d_bindings` which itself imports the mesh2d types needed for the bindings.
# Objective
Models can be produced that do not have vertex tangents but do have normal map textures. The tangents can be generated. There is a way that the vertex tangents can be generated to be exactly invertible to avoid introducing error when recreating the normals in the fragment shader.
## Solution
- After attempts to get https://github.com/gltf-rs/mikktspace to integrate simple glam changes and version bumps, and releases of that crate taking weeks / not being made (no offense intended to the authors/maintainers, bevy just has its own timelines and needs to take care of) it was decided to fork that repository. The following steps were taken:
- mikktspace was forked to https://github.com/bevyengine/mikktspace in order to preserve the repository's history in case the original is ever taken down
- The README in that repo was edited to add a note stating from where the repository was forked and explaining why
- The repo was locked for changes as its only purpose is historical
- The repo was integrated into the bevy repo using `git subtree add --prefix crates/bevy_mikktspace git@github.com:bevyengine/mikktspace.git master`
- In `bevy_mikktspace`:
- The travis configuration was removed
- `cargo fmt` was run
- The `Cargo.toml` was conformed to bevy's (just adding bevy to the keywords, changing the homepage and repository, changing the version to 0.7.0-dev - importantly the license is exactly the same)
- Remove the features, remove `nalgebra` entirely, only use `glam`, suppress clippy.
- This was necessary because our CI runs clippy with `--all-features` and the `nalgebra` and `glam` features are mutually exclusive, plus I don't want to modify this highly numerically-sensitive code just to appease clippy and diverge even more from upstream.
- Rebase https://github.com/bevyengine/bevy/pull/1795
- @jakobhellermann said it was fine to copy and paste but it ended up being almost exactly the same with just a couple of adjustments when validating correctness so I decided to actually rebase it and then build on top of it.
- Use the exact same fragment shader code to ensure correct normal mapping.
- Tested with both https://github.com/KhronosGroup/glTF-Sample-Models/tree/master/2.0/NormalTangentMirrorTest which has vertex tangents and https://github.com/KhronosGroup/glTF-Sample-Models/tree/master/2.0/NormalTangentTest which requires vertex tangent generation
Co-authored-by: alteous <alteous@outlook.com>
Adds ability to specify scaling factor for `WindowSize`, size of the fixed axis for `FixedVertical` and `FixedHorizontal` and a new `ScalingMode` that is a mix of `FixedVertical` and `FixedHorizontal`
# The issue
Currently, only available options are to:
* Have one of the axes fixed to value 1
* Have viewport size match the window size
* Manually adjust viewport size
In most of the games these options are not enough and more advanced scaling methods have to be used
## Solution
The solution is to provide additional parameters to current scaling modes, like scaling factor for `WindowSize`. Additionally, a more advanced `Auto` mode is added, which dynamically switches between behaving like `FixedVertical` and `FixedHorizontal` depending on the window's aspect ratio.
Co-authored-by: Daniikk1012 <49123959+Daniikk1012@users.noreply.github.com>
# Objective
allow meshes with equal z-depth to be rendered in a chosen order / avoid z-fighting
## Solution
add a depth_bias to SpecializedMaterial that is added to the mesh depth used for render-ordering.
# Objective
This PR aims to improve the soundness of `CommandQueue`. In particular it aims to:
- make it sound to store commands that contain padding or uninitialized bytes;
- avoid uses of commands after moving them in the queue's buffer (`std::mem::forget` is technically a use of its argument);
- remove useless checks: `self.bytes.as_mut_ptr().is_null()` is always `false` because even `Vec`s that haven't allocated use a dangling pointer. Moreover the same pointer was used to write the command, so it ought to be valid for reads if it was for writes.
## Solution
- To soundly store padding or uninitialized bytes `CommandQueue` was changed to contain a `Vec<MaybeUninit<u8>>` instead of `Vec<u8>`;
- To avoid uses of the command through `std::mem::forget`, `ManuallyDrop` was used.
## Other observations
While writing this PR I noticed that `CommandQueue` doesn't seem to drop the commands that weren't applied. While this is a pretty niche case (you would have to be manually using `CommandQueue`/`std::mem::swap`ping one), I wonder if it should be documented anyway.
# Objective
Don't allocate memory for Component types known at compile-time. Save a bit of memory.
## Solution
Change `ComponentDescriptor::name` from `String` to `Cow<'static, str>` to use the `&'static str` returned by `std::any::type_name`.
# Objective
`debug_assert!` macros must still compile properly in release mode due to how they're implemented. This is causing release builds to fail.
## Solution
Change them to `assert!` macros inside `#[cfg(debug_assertions)]` blocks.
# Objective
`bevy_reflect` as different kinds of reflected types (each with their own trait), `trait Struct: Reflect`, `trait List: Reflect`, `trait Map: Reflect`, ...
Types that don't fit either of those are called reflect value types, they are opaque and can't be deconstructed further.
`bevy_reflect` can serialize `dyn Reflect` values. Any container types (struct, list, map) get deconstructed and their elements serialized separately, which can all happen without serde being involved ever (happens [here](https://github.com/bevyengine/bevy/blob/main/crates/bevy_reflect/src/serde/ser.rs#L50-L85=)).
The only point at which we require types to be serde-serializable is for *value types* (happens [here](https://github.com/bevyengine/bevy/blob/main/crates/bevy_reflect/src/serde/ser.rs#L104=)).
So reflect array serializing is solved, since arrays are container types which don't require serde.
#1213 also introduced added the `serialize` method and `Serialize` impls for `dyn Array` and `DynamicArray` which use their element's `Reflect::serializable` function. This is 1. unnecessary, because it is not used for array serialization, and 2. annoying for removing the `Serialize` bound on container types, because these impls don't have access to the `TypeRegistry`, so we can't move the serialization code there.
# Solution
Remove these impls and `fn serialize`. It's not used and annoying for other changes.
# Objective
Increase compatibility with a fairly common format of padded spritesheets, in which half the padding value occurs before the first sprite box begins. The original behaviour falls out when `Vec2::ZERO` is used for `offset`.
See below unity screenshot for an example of a spritesheet with padding
![Screen Shot 2022-05-24 at 4 11 49 PM](https://user-images.githubusercontent.com/30442265/170123682-287e5733-b69d-452b-b2e6-46d8d29293fb.png)
## Solution
Tiny change to `crates/bevy_sprite/src/texture_atlas.rs`
## Migration Guide
Calls to `TextureAtlas::from_grid_with_padding` should be modified to include a new parameter, which can be set to `Vec2::ZERO` to retain old behaviour.
```rust
from_grid_with_padding(texture, tile_size, columns, rows, padding)
|
V
from_grid_with_padding(texture, tile_size, columns, rows, padding, Vec2::ZERO)
```
Co-authored-by: FraserLee <30442265+FraserLee@users.noreply.github.com>
# Objective
Currently, `FromReflect` makes a couple assumptions:
* Ignored fields must implement `Default`
* Active fields must implement `FromReflect`
* The reflected must be fully populated for active fields (can't use an empty `DynamicStruct`)
However, one or both of these requirements might be unachievable, such as for external types. In these cases, it might be nice to tell `FromReflect` to use a custom default.
## Solution
Added the `#[reflect(default)]` derive helper attribute. This attribute can be applied to any field (ignored or not) and will allow a default value to be specified in place of the regular `from_reflect()` call.
It takes two forms: `#[reflect(default)]` and `#[reflect(default = "some_func")]`. The former specifies that `Default::default()` should be used while the latter specifies that `some_func()` should be used. This is pretty much [how serde does it](https://serde.rs/field-attrs.html#default).
### Example
```rust
#[derive(Reflect, FromReflect)]
struct MyStruct {
// Use `Default::default()`
#[reflect(default)]
foo: String,
// Use `get_bar_default()`
#[reflect(default = "get_bar_default")]
#[reflect(ignore)]
bar: usize,
}
fn get_bar_default() -> usize {
123
}
```
### Active Fields
As an added benefit, this also allows active fields to be completely missing from their dynamic object. This is because the attribute tells `FromReflect` how to handle missing active fields (it still tries to use `from_reflect` first so the `FromReflect` trait is still required).
```rust
let dyn_struct = DynamicStruct::default();
// We can do this without actually including the active fields since they have `#[reflect(default)]`
let my_struct = <MyStruct as FromReflect>::from_reflect(&dyn_struct);
```
### Container Defaults
Also, with the addition of #3733, people will likely start adding `#[reflect(Default)]` to their types now. Just like with the fields, we can use this to mark the entire container as "defaultable". This grants us the ability to completely remove the field markers altogether if our type implements `Default` (and we're okay with fields using that instead of their own `Default` impls):
```rust
#[derive(Reflect, FromReflect)]
#[reflect(Default)]
struct MyStruct {
foo: String,
#[reflect(ignore)]
bar: usize,
}
impl Default for MyStruct {
fn default() -> Self {
Self {
foo: String::from("Hello"),
bar: 123,
}
}
}
// Again, we can now construct this from nothing pretty much
let dyn_struct = DynamicStruct::default();
let my_struct = <MyStruct as FromReflect>::from_reflect(&dyn_struct);
```
Now if _any_ field is missing when using `FromReflect`, we simply fallback onto the container's `Default` implementation.
This behavior can be completely overridden on a per-field basis, of course, by simply defining those same field attributes like before.
### Related
* #3733
* #1395
* #2377
---
## Changelog
* Added `#[reflect(default)]` field attribute for `FromReflect`
* Allows missing fields to be given a default value when using `FromReflect`
* `#[reflect(default)]` - Use the field's `Default` implementation
* `#[reflect(default = "some_fn")]` - Use a custom function to get the default value
* Allow `#[reflect(Default)]` to have a secondary usage as a container attribute
* Allows missing fields to be given a default value based on the container's `Default` impl when using `FromReflect`
Co-authored-by: Gino Valente <49806985+MrGVSV@users.noreply.github.com>
# Objective
- Add an `ExtractResourcePlugin` for convenience and consistency
## Solution
- Add an `ExtractResourcePlugin` similar to `ExtractComponentPlugin` but for ECS `Resource`s. The system that is executed simply clones the main world resource into a render world resource, if and only if the main world resource was either added or changed since the last execution of the system.
- Add an `ExtractResource` trait with a `fn extract_resource(res: &Self) -> Self` function. This is used by the `ExtractResourcePlugin` to extract the resource
- Add a derive macro for `ExtractResource` on a `Resource` with the `Clone` trait, that simply returns `res.clone()`
- Use `ExtractResourcePlugin` wherever both possible and appropriate
This was first done in 7b4e3a5, but was then reverted when the new
renderer for 0.6 was merged (ffecb05).
I'm assuming it was simply a mistake when merging.
# Objective
- Same as #2740, I think it was reverted by mistake when merging.
> # Objective
>
> - Make it easy to use HexColorError with `thiserror`, i.e. converting it into other error types.
>
> Makes this possible:
>
> ```rust
> #[derive(Debug, thiserror::Error)]
> pub enum LdtkError {
> #[error("An error occured while deserializing")]
> Json(#[from] serde_json::Error),
> #[error("An error occured while parsing a color")]
> HexColor(#[from] bevy::render::color::HexColorError),
> }
> ```
>
> ## Solution
>
> - Derive thiserror::Error the same way we do elsewhere (see query.rs for instance)
# Objective
- Higher order system could not be created by users.
- However, a simple change to `SystemParamFunction` allows this.
- Higher order systems in this case mean functions which return systems created using other systems, such as `chain` (which is basically equivalent to map)
## Solution
- Change `SystemParamFunction` to be a safe abstraction over `FnMut([In<In>,] ...params)->Out`.
- Note that I believe `SystemParamFunction` should not have been counted as part of our public api before this PR.
- This is because its only use was an unsafe function without an actionable safety comment.
- The safety comment was basically 'call this within bevy code'.
- I also believe that there are no external users in its current form.
- A quick search on Google and in the discord confirmed this.
## See also
- https://github.com/bevyengine/bevy/pull/4666, which uses this and subsumes the example here
---
## Changelog
### Added
- `SystemParamFunction`, which can be used to create higher order systems.
# Objective
Fixes#4353. Fixes#4431. Picks up fixes for a panic for `gilrs` when `getGamepads()` is not available.
## Solution
Update the `gilrs` to `v0.9.0`. Changelog can be seen here: dba36f9186
EDIT: Updated `uuid` to 1.1 to avoid duplicate dependencies. Added `nix`'s two dependencies as exceptions until `rodio` updates their deps.
# Objective
- Add Vertex Color support to 2D meshes and ColorMaterial. This extends the work from #4528 (which in turn builds on the excellent tangent handling).
## Solution
- Added `#ifdef` wrapped support for vertex colors in the 2D mesh shader and `ColorMaterial` shader.
- Added an example, `mesh2d_vertex_color_texture` to demonstrate it in action.
![image](https://user-images.githubusercontent.com/14896751/169530930-6ae0c6be-2f69-40e3-a600-ba91d7178bc3.png)
---
## Changelog
- Added optional (ifdef wrapped) vertex color support to the 2dmesh and color material systems.
# Objective
- Sometimes, people might load an asset as one type, then use it with an `Asset`s for a different type.
- See e.g. #4784.
- This is especially likely with the Gltf types, since users may not have a clear conceptual model of what types the assets will be.
- We had an instance of this ourselves, in the `scene_viewer` example
## Solution
- Make `Assets::get` require a type safe handle.
---
## Changelog
### Changed
- `Assets::<T>::get` and `Assets::<T>::get_mut` now require that the passed handles are `Handle<T>`, improving the type safety of handles.
### Added
- `HandleUntyped::typed_weak`, a helper function for creating a weak typed version of an exisitng `HandleUntyped`.
## Migration Guide
`Assets::<T>::get` and `Assets::<T>::get_mut` now require that the passed handles are `Handle<T>`, improving the type safety of handles. If you were previously passing in:
- a `HandleId`, use `&Handle::weak(id)` instead, to create a weak handle. You may have been able to store a type safe `Handle` instead.
- a `HandleUntyped`, use `&handle_untyped.typed_weak()` to create a weak handle of the specified type. This is most likely to be the useful when using [load_folder](https://docs.rs/bevy_asset/latest/bevy_asset/struct.AssetServer.html#method.load_folder)
- a `Handle<U>` of of a different type, consider whether this is the correct handle type to store. If it is (i.e. the same handle id is used for multiple different Asset types) use `Handle::weak(handle.id)` to cast to a different type.
# Objective
Fixes#4791. `ParallelExecutor` inserts a default `CompteTaskPool` if there isn't one stored as a resource, including when it runs on a different world. When spawning the render sub-app, the main world's `ComputeTaskPool` is not cloned and inserted into the render app's, which causes a second `ComputeTaskPool` with the default configuration to be spawned. This results in an excess number of threads being spawned.
## Solution
Copy the task pools from the main world to the subapps upon creating them.
## Alternative
An alternative to this would be to make the task pools global, as seen in #2250 or bevyengine/rfcs#54.
# Objective
Resolves#4753
## Solution
Using rust doc I added documentation to the struct. Decided to not provide an example in the doc comment but instead refer to the example file that shows the usage.
# Objective
Use less memory to store SparseSet components.
## Solution
Change `ComponentSparseSet` to only use `Entity::id` in it's key internally, and change the usize value in it's SparseArray to use u32 instead, as it cannot have more than u32::MAX live entities stored at once.
This should reduce the overhead of storing components in sparse set storage by 50%.
# Objective
Fixes#3183. Requiring a `&TaskPool` parameter is sort of meaningless if the only correct one is to use the one provided by `Res<ComputeTaskPool>` all the time.
## Solution
Have `QueryState` save a clone of the `ComputeTaskPool` which is used for all `par_for_each` functions.
~~Adds a small overhead of the internal `Arc` clone as a part of the startup, but the ergonomics win should be well worth this hardly-noticable overhead.~~
Updated the docs to note that it will panic the task pool is not present as a resource.
# Future Work
If https://github.com/bevyengine/rfcs/pull/54 is approved, we can replace these resource lookups with a static function call instead to get the `ComputeTaskPool`.
---
## Changelog
Removed: The `task_pool` parameter of `Query(State)::par_for_each(_mut)`. These calls will use the `World`'s `ComputeTaskPool` resource instead.
## Migration Guide
The `task_pool` parameter for `Query(State)::par_for_each(_mut)` has been removed. Remove these parameters from all calls to these functions.
Before:
```rust
fn parallel_system(
task_pool: Res<ComputeTaskPool>,
query: Query<&MyComponent>,
) {
query.par_for_each(&task_pool, 32, |comp| {
...
});
}
```
After:
```rust
fn parallel_system(query: Query<&MyComponent>) {
query.par_for_each(32, |comp| {
...
});
}
```
If using `Query(State)` outside of a system run by the scheduler, you may need to manually configure and initialize a `ComputeTaskPool` as a resource in the `World`.
# Objective
The `ComponentId` in `Column` is redundant as it's stored in parallel in the surrounding `SparseSet` all the time.
## Solution
Remove it. Add `SparseSet::iter(_mut)` to parallel `HashMap::iter(_mut)` to allow iterating pairs of columns and their IDs.
---
## Changelog
Added: `SparseSet::iter` and `SparseSet::iter_mut`.
# Objective
- Rebase of #3159.
- Fixes https://github.com/bevyengine/bevy/issues/3156
- add #[inline] to single related functions so that they matches with other function defs
## Solution
* added functions to QueryState
* get_single_unchecked_manual
* get_single_unchecked
* get_single
* get_single_mut
* single
* single_mut
* make Query::get_single use QueryState::get_single_unchecked_manual
* added #[inline]
---
## Changelog
### Added
Functions `QueryState::single`, `QueryState::get_single`, `QueryState::single_mut`, `QueryState::get_single_mut`, `QueryState::get_single_unchecked`, `QueryState::get_single_unchecked_manual`.
### Changed
`QuerySingleError` is now in the `state` module.
## Migration Guide
Change `query::QuerySingleError` to `state::QuerySingleError`
Co-authored-by: 2ne1ugly <chattermin@gmail.com>
Co-authored-by: 2ne1ugly <47616772+2ne1ugly@users.noreply.github.com>
# Objective
the code in these fns are always identical so stop having two functions
## Solution
make them the same function
---
## Changelog
change `matches_archetype` and `matches_table` to `fn matches_component_set(&self, &SparseArray<ComponentId, usize>) -> bool` then do extremely boring updating of all `FetchState` impls
## Migration Guide
- move logic of `matches_archetype` and `matches_table` into `matches_component_set` in any manual `FetchState` impls
# Objective
Debugging reflected types can be somewhat frustrating since all `dyn Reflect` trait objects return something like `Reflect(core::option::Option<alloc::string::String>)`.
It would be much nicer to be able to see the actual value— or even use a custom `Debug` implementation.
## Solution
Added `Reflect::debug` which allows users to customize the debug output. It sets defaults for all `ReflectRef` subtraits and falls back to `Reflect(type_name)` if no `Debug` implementation was registered.
To register a custom `Debug` impl, users can add `#[reflect(Debug)]` like they can with other traits.
### Example
Using the following structs:
```rust
#[derive(Reflect)]
pub struct Foo {
a: usize,
nested: Bar,
#[reflect(ignore)]
_ignored: NonReflectedValue,
}
#[derive(Reflect)]
pub struct Bar {
value: Vec2,
tuple_value: (i32, String),
list_value: Vec<usize>,
// We can't determine debug formatting for Option<T> yet
unknown_value: Option<String>,
custom_debug: CustomDebug
}
#[derive(Reflect)]
#[reflect(Debug)]
struct CustomDebug;
impl Debug for CustomDebug {
fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
write!(f, "This is a custom debug!")
}
}
pub struct NonReflectedValue {
_a: usize,
}
```
We can do:
```rust
let value = Foo {
a: 1,
_ignored: NonReflectedValue { _a: 10 },
nested: Bar {
value: Vec2::new(1.23, 3.21),
tuple_value: (123, String::from("Hello")),
list_value: vec![1, 2, 3],
unknown_value: Some(String::from("World")),
custom_debug: CustomDebug
},
};
let reflected_value: &dyn Reflect = &value;
println!("{:#?}", reflected_value)
```
Which results in:
```rust
Foo {
a: 2,
nested: Bar {
value: Vec2(
1.23,
3.21,
),
tuple_value: (
123,
"Hello",
),
list_value: [
1,
2,
3,
],
unknown_value: Reflect(core::option::Option<alloc::string::String>),
custom_debug: This is a custom debug!,
},
}
```
Notice that neither `Foo` nor `Bar` implement `Debug`, yet we can still deduce it. This might be a concern if we're worried about leaking internal values. If it is, we might want to consider a way to exclude fields (possibly with a `#[reflect(hide)]` macro) or make it purely opt in (as opposed to the default implementation automatically handled by ReflectRef subtraits).
Co-authored-by: Gino Valente <49806985+MrGVSV@users.noreply.github.com>
# Objective
Even if bevy itself does not provide any builtin scripting or modding APIs, it should have the foundations for building them yourself.
For that it should be enough to have APIs that are not tied to the actual rust types with generics, but rather accept `ComponentId`s and `bevy_ptr` ptrs.
## Solution
Add the following APIs to bevy
```rust
fn EntityRef::get_by_id(ComponentId) -> Option<Ptr<'w>>;
fn EntityMut::get_by_id(ComponentId) -> Option<Ptr<'_>>;
fn EntityMut::get_mut_by_id(ComponentId) -> Option<MutUntyped<'_>>;
fn World::get_resource_by_id(ComponentId) -> Option<Ptr<'_>>;
fn World::get_resource_mut_by_id(ComponentId) -> Option<MutUntyped<'_>>;
// Safety: `value` must point to a valid value of the component
unsafe fn World::insert_resource_by_id(ComponentId, value: OwningPtr);
fn ComponentDescriptor::new_with_layout(..) -> Self;
fn World::init_component_with_descriptor(ComponentDescriptor) -> ComponentId;
```
~~This PR would definitely benefit from #3001 (lifetime'd pointers) to make sure that the lifetimes of the pointers are valid and the my-move pointer in `insert_resource_by_id` could be an `OwningPtr`, but that can be adapter later if/when #3001 is merged.~~
### Not in this PR
- inserting components on entities (this is very tied to types with bundles and the `BundleInserter`)
- an untyped version of a query (needs good API design, has a large implementation complexity, can be done in a third-party crate)
Co-authored-by: Jakob Hellermann <hellermann@sipgate.de>
# Objective
One way to avoid texture atlas bleeding is to ensure that every vertex is
placed at an integer pixel coordinate. This is a particularly appealing
solution for regular structures like tile maps.
Doing so is currently harder than necessary when the WindowSize scaling
mode and Center origin are used: For odd window width or height, the
origin of the coordinate system is placed in the middle of a pixel at
some .5 offset.
## Solution
Avoid this issue by rounding the half width and height values.
Updates the requirements on [tracing-tracy](https://github.com/nagisa/rust_tracy_client) to permit the latest version.
<details>
<summary>Commits</summary>
<ul>
<li><a href="13b335a710"><code>13b335a</code></a> Remove ability to disable the client at runtime</li>
<li><a href="69e44977ee"><code>69e4497</code></a> The upgrades to 0.8.1</li>
<li><a href="c204b60c7a"><code>c204b60</code></a> Cancel the old test runs</li>
<li><a href="939bd04c1c"><code>939bd04</code></a> Remove the thread initialization calls</li>
<li><a href="7024e776bb"><code>7024e77</code></a> Update Tracy client bindings to v0.8.1</li>
<li><a href="5c54baa244"><code>5c54baa</code></a> tracy-client 0.12.7</li>
<li><a href="f183050b20"><code>f183050</code></a> Non-allocating <code>span!</code> macro</li>
<li><a href="15936ea751"><code>15936ea</code></a> tracy-client 0.12.6</li>
<li><a href="26d0c50542"><code>26d0c50</code></a> Relax literal the requirement of the create_plot macro so that it can be used...</li>
<li>See full diff in <a href="https://github.com/nagisa/rust_tracy_client/compare/tracing-tracy-v0.8.0...tracing-tracy-v0.9.0">compare view</a></li>
</ul>
</details>
<br />
Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.
[//]: # (dependabot-automerge-start)
[//]: # (dependabot-automerge-end)
---
<details>
<summary>Dependabot commands and options</summary>
<br />
You can trigger Dependabot actions by commenting on this PR:
- `@dependabot rebase` will rebase this PR
- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it
- `@dependabot merge` will merge this PR after your CI passes on it
- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it
- `@dependabot cancel merge` will cancel a previously requested merge and block automerging
- `@dependabot reopen` will reopen this PR if it is closed
- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually
- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)
</details>
# Objective
- Make bevy_app's optional bevy_reflect dependency actually optional
- Because bevy_ecs has a default dependency on bevy_reflect, bevy_app includes bevy_reflect transitively even with default-features=false, despite the optional dependency indicating that it was intended to be able to leave out bevy_reflect.
## Solution
- Make bevy_app not enable bevy_ecs's default features, and then use [the `dep:` syntax](https://doc.rust-lang.org/cargo/reference/features.html#optional-dependencies) introduced in 1.60 to make the default bevy_reflect feature enable bevy_ecs's bevy_reflect feature/dependency.
---
## Changelog
- bevy_app no longer enables bevy_ecs's `bevy_reflect` feature when included without its own `bevy_reflect` feature (which is on by default).
# Objective
Reduce the catch-all grab-bag of functionality in bevy_core by minimally splitting off time functionality into bevy_time. Functionality like that provided by #3002 would increase the complexity of bevy_time, so this is a good candidate for pulling into its own unit.
A step in addressing #2931 and splitting bevy_core into more specific locations.
## Solution
Pull the time module of bevy_core into a new crate, bevy_time.
# Migration guide
- Time related types (e.g. `Time`, `Timer`, `Stopwatch`, `FixedTimestep`, etc.) should be imported from `bevy::time::*` rather than `bevy::core::*`.
- If you were adding `CorePlugin` manually, you'll also want to add `TimePlugin` from `bevy::time`.
- The `bevy::core::CorePlugin::Time` system label is replaced with `bevy::time::TimeSystem`.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
- As noticed in #4333 by @x-52, the exact purpose and logic of `HasRawWIndowHandleWrapper` is unclear
- Unfortunately, there are rather good reasons why this design is needed (and why we can't just `impl HasRawWindowHandle for RawWindowHandleWrapper`
## Solution
- Rename `HasRawWindowHandleWrapper` to `ThreadLockedRawWindowHandleWrapper`, reflecting the primary distinction
- Document how this design is intended to be used
- Leave comments explaining why this design must exist
## Migration Guide
- renamed `HasRawWindowHandleWrapper` to `ThreadLockedRawWindowHandleWrapper`
# Objective
Make the function consistent with returned values and `as_hsla` method
Fixes#4826
## Solution
- Rename the method
## Migration Guide
- Rename the method
Currently Bevy's web canvases are "fixed size". They are manually set to specific dimensions. This might be fine for some games and website layouts, but for sites with flexible layouts, or games that want to "fill" the browser window, Bevy doesn't provide the tools needed to make this easy out of the box.
There are third party plugins like [bevy-web-resizer](https://github.com/frewsxcv/bevy-web-resizer/) that listen for window resizes, take the new dimensions, and resize the winit window accordingly. However this only covers a subset of cases and this is common enough functionality that it should be baked into Bevy.
A significant motivating use case here is the [Bevy WASM Examples page](https://bevyengine.org/examples/). This scales the canvas to fit smaller windows (such as mobile). But this approach both breaks winit's mouse events and removes pixel-perfect rendering (which means we might be rendering too many or too few pixels). https://github.com/bevyengine/bevy-website/issues/371
In an ideal world, winit would support this behavior out of the box. But unfortunately that seems blocked for now: https://github.com/rust-windowing/winit/pull/2074. And it builds on the ResizeObserver api, which isn't supported in all browsers yet (and is only supported in very new versions of the popular browsers).
While we wait for a complete winit solution, I've added a `fit_canvas_to_parent` option to WindowDescriptor / Window, which when enabled will listen for window resizes and resize the Bevy canvas/window to fit its parent element. This enables users to scale bevy canvases using arbitrary CSS, by "inheriting" their parents' size. Note that the wrapper element _is_ required because winit overrides the canvas sizing with absolute values on each resize.
There is one limitation worth calling out here: while the majority of canvas resizes will be triggered by window resizes, modifying element layout at runtime (css animations, javascript-driven element changes, dev-tool-injected changes, etc) will not be detected here. I'm not aware of a good / efficient event-driven way to do this outside of the ResizeObserver api. In practice, window-resize-driven canvas resizing should cover the majority of use cases. Users that want to actively poll for element resizes can just do that (or we can build another feature and let people choose based on their specific needs).
I also took the chance to make a couple of minor tweaks:
* Made the `canvas` window setting available on all platforms. Users shouldn't need to deal with cargo feature selection to support web scenarios. We can just ignore the value on non-web platforms. I added documentation that explains this.
* Removed the redundant "initial create windows" handler. With the addition of the code in this pr, the code duplication was untenable.
This enables a number of patterns:
## Easy "fullscreen window" mode for the default canvas
The "parent element" defaults to the `<body>` element.
```rust
app
.insert_resource(WindowDescriptor {
fit_canvas_to_parent: true,
..default()
})
```
And CSS:
```css
html, body {
margin: 0;
height: 100%;
}
```
## Fit custom canvas to "wrapper" parent element
```rust
app
.insert_resource(WindowDescriptor {
fit_canvas_to_parent: true,
canvas: Some("#bevy".to_string()),
..default()
})
```
And the HTML:
```html
<div style="width: 50%; height: 100%">
<canvas id="bevy"></canvas>
</div>
```
# Objective
Allow `Box<dyn Reflect>` to be converted into a `Box<dyn MyTrait>` using the `#[reflect_trait]` macro. The other methods `get` and `get_mut` only provide a reference to the reflected object.
## Solution
Add a `get_boxed` method to the `Reflect***` struct generated by the `#[reflect_trait]` macro. This method takes in a `Box<dyn Reflect>` and returns a `Box<dyn MyTrait>`.
Co-authored-by: MrGVSV <49806985+MrGVSV@users.noreply.github.com>
# Objective
Fixes#4657
Example code that wasnt panic'ing before this PR (and so was unsound):
```rust
#[test]
#[should_panic = "error[B0001]"]
fn option_has_no_filter_with() {
fn sys(_1: Query<(Option<&A>, &mut B)>, _2: Query<&mut B, Without<A>>) {}
let mut world = World::default();
run_system(&mut world, sys);
}
#[test]
#[should_panic = "error[B0001]"]
fn any_of_has_no_filter_with() {
fn sys(_1: Query<(AnyOf<(&A, ())>, &mut B)>, _2: Query<&mut B, Without<A>>) {}
let mut world = World::default();
run_system(&mut world, sys);
}
#[test]
#[should_panic = "error[B0001]"]
fn or_has_no_filter_with() {
fn sys(_1: Query<&mut B, Or<(With<A>, With<B>)>>, _2: Query<&mut B, Without<A>>) {}
let mut world = World::default();
run_system(&mut world, sys);
}
```
## Solution
- Only add the intersection of `with`/`without` accesses of all the elements in `Or/AnyOf` to the world query's `FilteredAccess<ComponentId>` instead of the union.
- `Option`'s fix can be thought of the same way since its basically `AnyOf<T, ()>` but its impl is just simpler as `()` has no `with`/`without` accesses
---
## Changelog
- `Or`/`AnyOf`/`Option` will now report more query conflicts in order to fix unsoundness
## Migration Guide
- If you are now getting query conflicts from `Or`/`AnyOf`/`Option` rip to you and ur welcome for it now being caught
# Objective
We have duplicated code between `QueryIter` and `QueryIterationCursor`. Reuse that code.
## Solution
- Reuse `QueryIterationCursor` inside `QueryIter`.
- Slim down `QueryIter` by removing the `&'w World`. It was only being used by the `size_hint` and `ExactSizeIterator` impls, which can use the QueryState and &Archetypes in the type already.
- Benchmark to make sure there is no significant regression.
Relevant benchmark results seem to show that there is no tangible difference between the two. Everything seems to be either identical or within a workable margin of error here.
```
group embed-cursor main
----- ------------ ----
fragmented_iter/base 1.00 387.4±19.70ns ? ?/sec 1.07 413.1±27.95ns ? ?/sec
many_maps_iter 1.00 27.3±0.22ms ? ?/sec 1.00 27.4±0.10ms ? ?/sec
simple_iter/base 1.00 13.8±0.07µs ? ?/sec 1.00 13.7±0.17µs ? ?/sec
simple_iter/sparse 1.00 61.9±0.37µs ? ?/sec 1.00 62.2±0.64µs ? ?/sec
simple_iter/system 1.00 13.7±0.34µs ? ?/sec 1.00 13.7±0.10µs ? ?/sec
sparse_fragmented_iter/base 1.00 11.0±0.54ns ? ?/sec 1.03 11.3±0.48ns ? ?/sec
world_query_iter/50000_entities_sparse 1.08 105.0±2.68µs ? ?/sec 1.00 97.5±2.18µs ? ?/sec
world_query_iter/50000_entities_table 1.00 27.3±0.13µs ? ?/sec 1.00 27.3±0.37µs ? ?/sec
```
# Objective
Quick followup to #4712.
While updating some [other PRs](https://github.com/bevyengine/bevy/pull/4218), I realized the `ReflectTraits` struct could be improved. The issue with the current implementation is that `ReflectTraits::get_xxx_impl(...)` returns just the _logic_ to the corresponding `Reflect` trait method, rather than the entire function.
This makes it slightly more annoying to manage since the variable names need to be consistent across files. For example, `get_partial_eq_impl` uses a `value` variable. But the name "value" isn't defined in the `get_partial_eq_impl` method, it's defined in three other methods in a completely separate file.
It's not likely to cause any bugs if we keep it as it is since differing variable names will probably just result in a compile error (except in very particular cases). But it would be useful to someone who wanted to edit/add/remove a method.
## Solution
Made `get_hash_impl`, `get_partial_eq_impl` and `get_serialize_impl` return the entire method implementation for `reflect_hash`, `reflect_partial_eq`, and `serializable`, respectively.
As a result of this, those three `Reflect` methods were also given default implementations. This was fairly simple to do since all three could just be made to return `None`.
---
## Changelog
* Small cleanup/refactor to `ReflectTraits` in `bevy_reflect_derive`
* Gave `Reflect::reflect_hash`, `Reflect::reflect_partial_eq`, and `Reflect::serializable` default implementations
# Objective
Support returning data out of with_children to enable the use case of changing the parent commands with data created inside the child builder.
## Solution
Change the with_children closure to return T.
Closes https://github.com/bevyengine/bevy/pull/2817.
---
## Changelog
`BuildChildren::add_children` was added with the ability to return data to use outside the closure (for spawning a new child builder on a returned entity for example).
# Objective
- We do a lot of function pointer calls in a hot loop (clearing entities in render). This is slow, since calling function pointers cannot be optimised out. We can avoid that in the cases where the function call is a no-op.
- Alternative to https://github.com/bevyengine/bevy/pull/2897
- On my machine, in `many_cubes`, this reduces dropping time from ~150μs to ~80μs.
## Solution
- Make `drop` in `BlobVec` an `Option`, recording whether the given drop impl is required or not.
- Note that this does add branching in some cases - we could consider splitting this into two fields, i.e. unconditionally call the `drop` fn pointer.
- My intuition of how often types stored in `World` should have non-trivial drops makes me think that would be slower, however.
N.B. Even once this lands, we should still test having a 'drop_multiple' variant - for types with a real `Drop` impl, the current implementation is definitely optimal.
# Objective
- Part of the splitting process of #3692.
## Solution
- Document `keyboard.rs` inside of `bevy_input`.
Co-authored-by: KDecay <KDecayMusic@protonmail.com>
# Objective
- Transform propogation could stack overflow when there was a cycle.
- I think https://github.com/bevyengine/bevy/pull/4203 would use all available memory.
## Solution
- Make sure that the child entity's `Parent`s are their parents.
This is also required for when parallelising, although as noted in the comment, the naïve solution would be UB.
(The best way to fix this would probably be an `&mut UnsafeCell<T>` `WorldQuery`, or wrapper type with the same effect)
# Objective
`bevy_ptr` works just fine without `std`. Mark it as `no_std`. This should generally be useful for non-bevy use cases, but it also marginally speeds up compilation by allowing the crate to compile without loading the std-lib.
## Solution
Replace `std` with `core`. Added `#![no_std]` to the crate and to the crate's tags.
Also added a missing `#![warn(missing_docs)]` that the other crates have.
# Objective
- Fixes#4456
## Solution
- Removed the `near` and `far` fields from the camera and the views.
---
## Changelog
- Removed the `near` and `far` fields from the camera and the views.
- Removed the `ClusterFarZMode::CameraFarPlane` far z mode.
## Migration Guide
- Cameras no longer accept near and far values during initialization
- `ClusterFarZMode::Constant` should be used with the far value instead of `ClusterFarZMode::CameraFarPlane`
# Objective
`bevy_ecs` assumes that `u32 as usize` is a lossless operation and in a few cases relies on this for soundness and correctness. The only platforms that Rust compiles to where this invariant is broken are 16-bit systems.
A very clear example of this behavior is in the SparseSetIndex impl for Entity, where it converts a u32 into a usize to act as an index. If usize is 16-bit, the conversion will overflow and provide the caller with the wrong index. This can easily result in previously unforseen aliased mutable borrows (i.e. Query::get_many_mut).
## Solution
Explicitly fail compilation on 16-bit platforms instead of introducing UB.
Properly supporting 16-bit systems will likely need a workable use case first.
---
## Changelog
Removed: Ability to compile `bevy_ecs` on 16-bit platforms.
## Migration Guide
`bevy_ecs` will now explicitly fail to compile on 16-bit platforms. If this is required, there is currently no alternative. Please file an issue (https://github.com/bevyengine/bevy/issues) to help detail your use case.
# Objective
> ℹ️ **Note**: This is a rebased version of #2383. A large portion of it has not been touched (only a few minor changes) so that any additional discussion may happen here. All credit should go to @NathanSWard for their work on the original PR.
- Currently reflection is not supported for arrays.
- Fixes#1213
## Solution
* Implement reflection for arrays via the `Array` trait.
* Note, `Array` is different from `List` in the way that you cannot push elements onto an array as they are statically sized.
* Now `List` is defined as a sub-trait of `Array`.
---
## Changelog
* Added the `Array` reflection trait
* Allows arrays up to length 32 to be reflected via the `Array` trait
## Migration Guide
* The `List` trait now has the `Array` supertrait. This means that `clone_dynamic` will need to specify which version to use:
```rust
// Before
let cloned = my_list.clone_dynamic();
// After
let cloned = List::clone_dynamic(&my_list);
```
* All implementers of `List` will now need to implement `Array` (this mostly involves moving the existing methods to the `Array` impl)
Co-authored-by: NathanW <nathansward@comcast.net>
Co-authored-by: MrGVSV <49806985+MrGVSV@users.noreply.github.com>
# Objective
- It's pretty common to want to check if an EventReader has received one or multiple events while also needing to consume the iterator to "clear" the EventReader.
- The current approach is to do something like `events.iter().count() > 0` or `events.iter().last().is_some()`. It's not immediately obvious that the purpose of that is to consume the events and check if there were any events. My solution doesn't really solve that part, but it encapsulates the pattern.
## Solution
- Add a `.clear()` method that consumes the iterator.
- It takes the EventReader by value to make sure it isn't used again after it has been called.
---
## Migration Guide
Not a breaking change, but if you ever found yourself in a situation where you needed to consume the EventReader and check if there was any events you can now use
```rust
fn system(events: EventReader<MyEvent>) {
if !events.is_empty {
events.clear();
// Process the fact that one or more event was received
}
}
```
Co-authored-by: Charles <IceSentry@users.noreply.github.com>
# Objective
`Query::par_for_each` and it's variants do not show up when profiling using `tracy` or other profilers. Failing to show the impact of changing batch size, the overhead of scheduling tasks, overall thread utilization, etc. other than the effect on the surrounding system.
## Solution
Add a child span that is entered on every spawned task.
Example view of the results in `tracy` using a modified `parallel_query`:
![image](https://user-images.githubusercontent.com/3137680/167560036-626bd091-344b-4664-b323-b692f4f16084.png)
---
## Changelog
Added: `tracing` spans for `Query::par_for_each` and its variants. Spans should now be visible for all
# Objective
The `bevy_reflect_derive` crate is not the cleanest or easiest to follow/maintain. The `lib.rs` file is especially difficult with over 1000 lines of code written in a confusing order. This is just a result of growth within the crate and it would be nice to clean it up for future work.
## Solution
Split `bevy_reflect_derive` into many more submodules. The submodules include:
* `container_attributes` - Code relating to container attributes
* `derive_data` - Code relating to reflection-based derive metadata
* `field_attributes` - Code relating to field attributes
* `impls` - Code containing actual reflection implementations
* `reflect_value` - Code relating to reflection-based value metadata
* `registration` - Code relating to type registration
* `utility` - General-purpose utility functions
This leaves the `lib.rs` file to contain only the public macros, making it much easier to digest (and fewer than 200 lines).
By breaking up the code into smaller modules, we make it easier for future contributors to find the code they're looking for or identify which module best fits their own additions.
### Metadata Structs
This cleanup also adds two big metadata structs: `ReflectFieldAttr` and `ReflectDeriveData`. The former is used to store all attributes for a struct field (if any). The latter is used to store all metadata for struct-based derive inputs.
Both significantly reduce code duplication and make editing these macros much simpler. The tradeoff is that we may collect more metadata than needed. However, this is usually a small thing (such as checking for attributes when they're not really needed or creating a `ReflectFieldAttr` for every field regardless of whether they actually have an attribute).
We could try to remove these tradeoffs and squeeze some more performance out, but doing so might come at the cost of developer experience. Personally, I think it's much nicer to create a `ReflectFieldAttr` for every field since it means I don't have to do two `Option` checks. Others may disagree, though, and so we can discuss changing this either in this PR or in a future one.
### Out of Scope
_Some_ documentation has been added or improved, but ultimately good docs are probably best saved for a dedicated PR.
## 🔍 Focus Points (for reviewers)
I know it's a lot to sift through, so here is a list of **key points for reviewers**:
- The following files contain code that was mostly just relocated:
- `reflect_value.rs`
- `registration.rs`
- `container_attributes.rs` was also mostly moved but features some general cleanup (reducing nesting, removing hardcoded strings, etc.) and lots of doc comments
- Most impl logic was moved from `lib.rs` to `impls.rs`, but they have been significantly modified to use the new `ReflectDeriveData` metadata struct in order to reduce duplication.
- `derive_data.rs` and `field_attributes.rs` contain almost entirely new code and should probably be given the most attention.
- Likewise, `from_reflect.rs` saw major changes using `ReflectDeriveData` so it should also be given focus.
- There was no change to the `lib.rs` exports so the end-user API should be the same.
## Prior Work
This task was initially tackled by @NathanSWard in #2377 (which was closed in favor of this PR), so hats off to them for beating me to the punch by nearly a year!
---
## Changelog
* **[INTERNAL]** Split `bevy_reflect_derive` into smaller submodules
* **[INTERNAL]** Add `ReflectFieldAttr`
* **[INTERNAL]** Add `ReflectDeriveData`
* Add `BevyManifest::get_path_direct()` method (`bevy_macro_utils`)
Co-authored-by: MrGVSV <49806985+MrGVSV@users.noreply.github.com>
# Objective
The frame marker event was emitted in the loop of presenting all the windows. This would mark the frame as finished multiple times if more than one window is used.
## Solution
Move the frame marker to after the `for`-loop, so that it gets executed only once.
# Objective
- The code in `events.rs` was a bit messy. There was lots of duplication between `EventReader` and `ManualEventReader`, and the state management code is not needed.
## Solution
- Clean it up.
## Future work
Should we remove the type parameter from `ManualEventReader`?
It doesn't have any meaning outside of its source `Events`. But there's no real reason why it needs to have a type parameter - it's just plain data. I didn't remove it yet to keep the type safety in some of the users of it (primarily related to `&mut World` usage)
# Objective
Relevant issue: #4474
Currently glam types implement Reflect as a value, which is problematic for reflection, making scripting/editor work much more difficult. This PR re-implements them as structs.
## Solution
Added a new proc macro, `impl_reflect_struct`, which replaces `impl_reflect_value` and `impl_from_reflect_value` for glam types. This macro could also be used for other types, but I don't know of any that would require it. It's specifically useful for foreign types that cannot derive Reflect normally.
---
## Changelog
### Added
- `impl_reflect_struct` proc macro
### Changed
- Glam reflect impls have been replaced with `impl_reflect_struct`
- from_reflect's `impl_struct` altered to take an optional custom constructor, allowing non-default non-constructible foreign types to use it
- Calls to `impl_struct` altered to conform to new signature
- Altered glam types (All vec/mat combinations) have a different serialization structure, as they are reflected differently now.
## Migration Guide
This will break altered glam types serialized to RON scenes, as they will expect to be serialized/deserialized as structs rather than values now. A future PR to add custom serialization for non-value types is likely on the way to restore previous behavior. Additionally, calls to `impl_struct` must add a `None` parameter to the end of the call to restore previous behavior.
Co-authored-by: PROMETHIA-27 <42193387+PROMETHIA-27@users.noreply.github.com>
Required for https://github.com/bevyengine/bevy/pull/4402.
# Objective
- derived `SystemParam` implementations were never `ReadOnlySystemParamFetch`
- We want them to be, e.g. for `EventReader`
## Solution
- If possible, 'forward' the impl of `ReadOnlySystemParamFetch`.
# Objective
- (Eventually) reduce noise in reporting access conflicts between unordered systems.
- `SystemStage` only looks at unfiltered `ComponentId` access, any conflicts reported are potentially `false`.
- the systems could still be accessing disjoint archetypes
- Comparing systems' filtered access sets can maybe avoid that (for statically known component types).
- #4204
## Solution
- Modify `SparseSetIndex` trait to require `PartialEq`, `Eq`, and `Hash` (all internal types except `BundleId` already did).
- Add `is_compatible` and `get_conflicts` methods to `FilteredAccessSet<T>`
- (existing method renamed to `get_conflicts_single`)
- Add docs for those and all the other methods while I'm at it.
## Objective
- ~~Make absurdly long-lived changes stay detectable for even longer (without leveling up to `u64`).~~
- Give all changes a consistent maximum lifespan.
- Improve code clarity.
## Solution
- ~~Increase the frequency of `check_tick` scans to increase the oldest reliably-detectable change.~~
(Deferred until we can benchmark the cost of a scan.)
- Ignore changes older than the maximum reliably-detectable age.
- General refactoring—name the constants, use them everywhere, and update the docs.
- Update test cases to check for the specified behavior.
## Related
This PR addresses (at least partially) the concerns raised in:
- #3071
- #3082 (and associated PR #3084)
## Background
- #1471
Given the minimum interval between `check_ticks` scans, `N`, the oldest reliably-detectable change is `u32::MAX - (2 * N - 1)` (or `MAX_CHANGE_AGE`). Reducing `N` from ~530 million (current value) to something like ~2 million would extend the lifetime of changes by a billion.
| minimum `check_ticks` interval | oldest reliably-detectable change | usable % of `u32::MAX` |
| --- | --- | --- |
| `u32::MAX / 8` (536,870,911) | `(u32::MAX / 4) * 3` | 75.0% |
| `2_000_000` | `u32::MAX - 3_999_999` | 99.9% |
Similarly, changes are still allowed to be between `MAX_CHANGE_AGE`-old and `u32::MAX`-old in the interim between `check_tick` scans. While we prevent their age from overflowing, the test to detect changes still compares raw values. This makes failure ultimately unreliable, since when ancient changes stop being detected varies depending on when the next scan occurs.
## Open Question
Currently, systems and system states are incorrectly initialized with their `last_change_tick` set to `0`, which doesn't handle wraparound correctly.
For consistent behavior, they should either be initialized to the world's `last_change_tick` (and detect no changes) or to `MAX_CHANGE_AGE` behind the world's current `change_tick` (and detect everything as a change). I've currently gone with the latter since that was closer to the existing behavior.
## Follow-up Work
(Edited: entire section)
We haven't actually profiled how long a `check_ticks` scan takes on a "large" `World` , so we don't know if it's safe to increase their frequency. However, we are currently relying on play sessions not lasting long enough to trigger a scan and apps not having enough entities/archetypes for it to be "expensive" (our assumption). That isn't a real solution. (Either scanning never costs enough to impact frame times or we provide an option to use `u64` change ticks. Nobody will accept random hiccups.)
To further extend the lifetime of changes, we actually only need to increment the world tick if a system has `Fetch: !ReadOnlySystemParamFetch`. The behavior will be identical because all writes are sequenced, but I'm not sure how to implement that in a way that the compiler can optimize the branch out.
Also, since having no false positives depends on a `check_ticks` scan running at least every `2 * N - 1` ticks, a `last_check_tick` should also be stored in the `World` so that any lull in system execution (like a command flush) could trigger a scan if needed. To be completely robust, all the systems initialized on the world should be scanned, not just those in the current stage.
# Objective
- Add two missing ogg vorbis audio extensions.
## Solution
- Add the two missing extensions to the list
- The file format is the same, there are simply two other possible extensions files can use.
- This can be easily (manually) tested by renaming the extension of `assets/sounds/Windless Slopes.ogg` to end in either `.oga` or `.spx` (in both the filesystem and in `examples/audio/audio.rs`) and then running `cargo run --example audio` and observing that the music still plays.
## More info
From the [wikipedia article for Ogg](https://en.wikipedia.org/wiki/Ogg):
> Ogg audio media is registered as [IANA](https://en.wikipedia.org/wiki/Internet_Assigned_Numbers_Authority) [media type](https://en.wikipedia.org/wiki/Media_type) audio/ogg with file extensions .oga, .ogg, and [.spx](https://en.wikipedia.org/wiki/Speex).
The current workaround is to rename any files ending in `.oga` or `.spx` to end in `.ogg` instead, which complicates tracking assets procured from other organizations.
See also [a corresponding change to bevy_kira_audio](https://github.com/NiklasEi/bevy_kira_audio/pull/8)
---
## Changelog
### Added
- Vorbis audio files may now use the `.oga` and `.spx` filename extensions in addition to the more common `.ogg` filename extension.
# Objective
It is possible to get a mutable reference to a `TypeRegistration` using
`TypeRegistry::get_mut`. However, none of its other methods
(`get_mut_with_name`, `get_type_data`, `iter`, etc.) have mutable
versions.
Besides improving consistency, this change would facilitate use cases
which involve storing mutable state data in the `TypeRegistry`.
## Solution
Provides a trivial wrapper around the mutable accessors that the
`TypeRegistration` already provides. Exactly mirrors the existing
immutable versions.
# Objective
Make it easy to get position and index data from Meshes.
## Solution
It was previously possible to get the mesh data by manually matching on `Mesh::VertexAttributeValues` and `Mesh::Indices`as in the bodies of these two methods (`VertexAttributeValues::as_float3(&self)` and `Indices::iter(&self)`), but that's needless duplication that making these methods `pub` fixes.
# Objective
- Remove `Resource` binding on events, introduce a new `Event` trait
- Ensure event iterators are `ExactSizeIterator`
## Solution
- Builds on #2382 and #2969
## Changelog
- Events<T>, EventWriter<T>, EventReader<T> and so on now require that the underlying type is Event, rather than Resource. Both of these are trivial supertraits of Send + Sync + 'static with universal blanket implementations: this change is currently purely cosmetic.
- Event reader iterators now implement ExactSizeIterator
In a `PluginGroupBuilder`, when adding a plugin that was already in the group (potentially disabled), it was then added twice to the app builder when calling `finish`. As the plugin is kept in an `HashMap`, it is not possible to have the same plugins twice with different configuration.
This PR updates the order of the plugin group so that each plugin is present only once.
Co-authored-by: François <8672791+mockersf@users.noreply.github.com>
# Objective
- noticed a few Vec3 and Vec2 that could be const
## Solution
- Declared them as const
- It seems to make a tiny improvement in example `many_light`, but given that the change is not complex at all it could still be worth it
# Objective
We have some macros that are public but only used internally for now. They fail on user's code due to the use of crate names like `bevy_utils`, while the user only has `bevy::utils`. There are two affected macros.
- `bevy_utils::define_label`: it may be useful in user's code for defining custom kinds of label traits (this is why I made this PR).
- `bevy_asset::load_internal_asset`: not useful currently due to limitations of the debug asset server, but this may change in the future.
## Solution
We can make them work by using `$crate` instead of names of their own crates, which can refer to the macro's defining crate regardless of the user's setup. Even though our objective is rather low-priority here, the solution adds no maintenance cost so it is still worthwhile.
1. change `PtrMut::as_ptr(self)` and `OwnedPtr::as_ptr(self)` to take `&self`, otherwise printing the pointer will prevent doing anything else afterwards
2. make all `as_ptr` methods safe. There's nothing unsafe about obtaining a pointer, these kinds of methods are safe in std as well [str::as_ptr](https://doc.rust-lang.org/stable/std/primitive.str.html#method.as_ptr), [Rc::as_ptr](https://doc.rust-lang.org/stable/std/rc/struct.Rc.html#method.as_ptr)
3. rename `offset`/`add` to `byte_offset`/`byte_add`. The unprefixed methods in std add in increments of `std::mem::size_of::<T>`, not in bytes. There's a PR for rust to add these byte_ methods https://github.com/rust-lang/rust/pull/95643 and at the call site it makes it much more clear that you need to do `.byte_add(i * layout_size)` instead of `.add(i)`
# Objective
Fixes#3180, builds from https://github.com/bevyengine/bevy/pull/2898
## Solution
Support requesting a window to be closed and closing a window in `bevy_window`, and handle this in `bevy_winit`.
This is a stopgap until we move to windows as entites, which I'm sure I'll get around to eventually.
## Changelog
### Added
- `Window::close` to allow closing windows.
- `WindowClosed` to allow reacting to windows being closed.
### Changed
Replaced `bevy::system::exit_on_esc_system` with `bevy:🪟:close_on_esc`.
## Fixed
The app no longer exits when any window is closed. This difference is only observable when there are multiple windows.
## Migration Guide
`bevy::input::system::exit_on_esc_system` has been removed. Use `bevy:🪟:close_on_esc` instead.
`CloseWindow` has been removed. Use `Window::close` instead.
The `Close` variant has been added to `WindowCommand`. Handle this by closing the relevant window.
# Objective
Fixes#4556
## Solution
StorageBuffer must use the Size of the std430 representation to calculate the buffer size, as the std430 representation is the data that will be written to it.
# Objective
Add support for vertex colors
## Solution
This change is modeled after how vertex tangents are handled, so the shader is conditionally compiled with vertex color support if the mesh has the corresponding attribute set.
Vertex colors are multiplied by the base color. I'm not sure if this is the best for all cases, but may be useful for modifying vertex colors without creating a new mesh.
I chose `VertexFormat::Float32x4`, but I'd prefer 16-bit floats if/when support is added.
## Changelog
### Added
- Vertex colors can be specified using the `Mesh::ATTRIBUTE_COLOR` mesh attribute.
# Objective
Bevy users often want to create circles and other simple shapes.
All the machinery is in place to accomplish this, and there are external crates that help. But when writing code for e.g. a new bevy example, it's not really possible to draw a circle without bringing in a new asset, writing a bunch of scary looking mesh code, or adding a dependency.
In particular, this PR was inspired by this interaction in another PR: https://github.com/bevyengine/bevy/pull/3721#issuecomment-1016774535
## Solution
This PR adds `shape::RegularPolygon` and `shape::Circle` (which is just a `RegularPolygon` that defaults to a large number of sides)
## Discussion
There's a lot of ongoing discussion about shapes in <https://github.com/bevyengine/rfcs/pull/12> and at least one other lingering shape PR (although it seems incomplete).
That RFC currently includes `RegularPolygon` and `Circle` shapes, so I don't think that having working mesh generation code in the engine for those shapes would add much burden to an author of an implementation.
But if we'd prefer not to add additional shapes until after that's sorted out, I'm happy to close this for now.
## Alternatives for users
For any users stumbling on this issue, here are some plugins that will help if you need more shapes.
https://github.com/Nilirad/bevy_prototype_lyonhttps://github.com/johanhelsing/bevy_smudhttps://github.com/Weasy666/bevy_svghttps://github.com/redpandamonium/bevy_more_shapeshttps://github.com/ForesightMiningSoftwareCorporation/bevy_polyline
# Objective
- When spawning a sprite the alpha is used for transparency, but when using the `Color::into()` implementation to spawn a `StandardMaterial`, the alpha is ignored.
- Pretty much everytime I want to make something transparent I started with a `Color::rgb().into()` and I'm always surprised that it doesn't work when changing it to `Color::rgba().into()`
- It's possible there's an issue with this approach I am not thinking of, but I'm not sure what's the point of setting an alpha value without the goal of making a color transparent.
## Solution
- Set the alpha_mode to AlphaMode::Blend when the alpha is not the default value.
---
## Migration Guide
This is not a breaking change, but it can easily be migrated to reduce boilerplate
```rust
commands.spawn_bundle(PbrBundle {
mesh: meshes.add(shape::Cube::default().into()),
material: materials.add(StandardMaterial {
base_color: Color::rgba(1.0, 0.0, 0.0, 0.75),
alpha_mode: AlphaMode::Blend,
..default()
}),
..default()
});
// becomes
commands.spawn_bundle(PbrBundle {
mesh: meshes.add(shape::Cube::default().into()),
material: materials.add(Color::rgba(1.0, 0.0, 0.0, 0.75).into()),
..default()
});
```
Co-authored-by: Charles <IceSentry@users.noreply.github.com>
# Objective
The pointer types introduced in #3001 are useful not just in `bevy_ecs`, but also in crates like `bevy_reflect` (#4475) or even outside of bevy.
## Solution
Extract `Ptr<'a>`, `PtrMut<'a>`, `OwnedPtr<'a>`, `ThinSlicePtr<'a, T>` and `UnsafeCellDeref` from `bevy_ecs::ptr` into `bevy_ptr`.
**Note:** `bevy_ecs` still reexports the `bevy_ptr` as `bevy_ecs::ptr` so that crates like `bevy_transform` can use the `Bundle` derive without needing to depend on `bevy_ptr` themselves.
# Objective
- `RunOnce` was a manual `System` implementation.
- Adding run criteria to stages was yet to be systemyoten
## Solution
- Make it a normal function
- yeet
## Changelog
- Replaced `RunOnce` with `ShouldRun::once`
## Migration guide
The run criterion `RunOnce`, which would make the controlled systems run only once, has been replaced with a new run criterion function `ShouldRun::once`. Replace all instances of `RunOnce` with `ShouldRun::once`.
# Objective
The `Ptr` types gives free access to the underlying `NonNull<u8>`, which adds more publicly visible pointer wrangling than there needs to be. There are also a few edge cases where Ptr types could be more readily utilized for properly validating the soundness of ECS operations.
## Solution
- Replace `*Ptr(Mut)::inner` with `cast` which requires a concrete type to give the pointer. This function could also have a `debug_assert` with an alignment check to ensure that the pointer is aligned properly, but is currently not included.
- Use `OwningPtr::read` in ECS macros over casting the inner pointer around.
# Objective
- After #3412, `Camera::world_to_screen` got a little bit uglier to use by needing to provide both `Windows` and `Assets<Image>`, even though only one would be needed b697e73c3d/crates/bevy_render/src/camera/camera.rs (L117-L123)
- Some time, exact coordinates are not needed but normalized device coordinates is enough
## Solution
- Add a function to just get NDC
### Problem
It currently isn't possible to construct the default value of a reflected type. Because of that, it isn't possible to use `add_component` of `ReflectComponent` to add a new component to an entity because you can't know what the initial value should be.
### Solution
1. add `ReflectDefault` type
```rust
#[derive(Clone)]
pub struct ReflectDefault {
default: fn() -> Box<dyn Reflect>,
}
impl ReflectDefault {
pub fn default(&self) -> Box<dyn Reflect> {
(self.default)()
}
}
impl<T: Reflect + Default> FromType<T> for ReflectDefault {
fn from_type() -> Self {
ReflectDefault {
default: || Box::new(T::default()),
}
}
}
```
2. add `#[reflect(Default)]` to all component types that implement `Default` and are user facing (so not `ComputedSize`, `CubemapVisibleEntities` etc.)
This makes it possible to add the default value of a component to an entity without any compile-time information:
```rust
fn main() {
let mut app = App::new();
app.register_type::<Camera>();
let type_registry = app.world.get_resource::<TypeRegistry>().unwrap();
let type_registry = type_registry.read();
let camera_registration = type_registry.get(std::any::TypeId::of::<Camera>()).unwrap();
let reflect_default = camera_registration.data::<ReflectDefault>().unwrap();
let reflect_component = camera_registration
.data::<ReflectComponent>()
.unwrap()
.clone();
let default = reflect_default.default();
drop(type_registry);
let entity = app.world.spawn().id();
reflect_component.add_component(&mut app.world, entity, &*default);
let camera = app.world.entity(entity).get::<Camera>().unwrap();
dbg!(&camera);
}
```
### Open questions
- should we have `ReflectDefault` or `ReflectFromWorld` or both?
# Objective
- While optimising many_cubes, I noticed that all material handles are extracted regardless of whether the entity to which the handle belongs is visible or not. As such >100k handles are extracted when only <20k are visible.
## Solution
- Only extract material handles of visible entities.
- This improves `many_cubes -- sphere` from ~42fps to ~48fps. It reduces not only the extraction time but also system commands time. `Handle<StandardMaterial>` extraction and its system commands went from 0.522ms + 3.710ms respectively, to 0.267ms + 0.227ms an 88% reduction for this system for this case. It's very view dependent but...
# Objective
- Creating and executing render passes has GPU overhead. If there are no phase items in the render phase to draw, then this overhead should not be incurred as it has no benefit.
## Solution
- Check if there are no phase items to draw, and if not, do not construct not execute the render pass
---
## Changelog
- Changed: Do not create nor execute empty render passes
# Objective
1. Previously, the `change_tick` and `last_change_tick` fields on `SystemChangeTick` [were `pub`](https://docs.rs/bevy/0.6.1/bevy/ecs/system/struct.SystemChangeTick.html).
1. This was actively misleading, as while this can be fetched as a `SystemParam`, a copy is returned instead
2. This information could be useful for debugging, but there was no way to investigate when data was changed.
3. There were no docs!
## Solution
1. Move these to a getter method.
2. Add `last_changed` method to the `DetectChanges` trait to enable inspection of when data was last changed.
3. Add docs.
# Changelog
`SystemChangeTick` now provides getter methods for the current and previous change tick, rather than public fields.
This can be combined with `DetectChanges::last_changed()` to debug the timing of changes.
# Migration guide
The `change_tick` and `last_change_tick` fields on `SystemChangeTick` are now private, use the corresponding getter method instead.
# Objective
avoid naming collisions with user structs when deriving ``system_param``.
## Solution
~rename the fetch struct created by ``#[derive(system_param)]`` from ``{}State`` to ``{}SysParamState``.~
place the fetch struct into an anonymous scope.
## Migration Guide
For code that was using a system param's fetch struct, such as ``EventReader``'s ``EventReaderState``, the fetch struct can now be identified via the SystemParam trait associated type ``Fetch``, e.g. for ``EventReader<T>`` it can be identified as ``<EventReader<'static, 'static, T> as SystemParam>::Fetch``
Supercedes https://github.com/bevyengine/bevy/pull/3340, and absorbs the test from there.
# Objective
- Fixes#3329
## Solution
- If the `Children` component has changed, we currently do not have a way to know how it has changed.
- Therefore, we must update the hierarchy downwards from that point to be correct.
Co-authored-by: Daniel McNab <36049421+DJMcNab@users.noreply.github.com>