# Objective
Fixes#6891
## Solution
Replaces deserializing map keys as `&str` with deserializing them as `String`.
This bug seems to occur when using something like `File` or `BufReader` rather than bytes or a string directly (I only tested `File` and `BufReader` for `rmp-serde` and `serde_json`). This might be an issue with other `Read` impls as well (except `&[u8]` it seems).
We already had passing tests for Message Pack but none that use a `File` or `BufReader`. This PR also adds or modifies tests to check for this in the future.
This change was also based on [feedback](https://github.com/bevyengine/bevy/pull/4561#discussion_r957385136) I received in a previous PR.
---
## Changelog
- Fix bug where scene deserialization using certain readers could fail (e.g. `BufReader`, `File`, etc.)
# Objective
Speed up animation by leveraging all threads in `ComputeTaskPool`.
## Solution
This PR parallelizes animation sampling across all threads.
To ensure that this is safely done, all animation is predicated with an ancestor query to ensure that there is no conflicting `AnimationPlayer` above each animated hierarchy that may cause this to alias.
Unlike the RFC, this does not add support for reflect based "animate anything", but only extends the existing `AnimationPlayer` to support high numbers of animated characters on screen at once.
## Performance
This cuts `many_foxes`'s frame time on my machine by a full millisecond, from 7.49ms to 6.5ms. (yellow is this PR, red is main).
---
## Changelog
Changed: Animation sampling now runs fully multi-threaded using threads from `ComputeTaskPool`.
Changed: `AnimationPlayer` that are on a child or descendant of another entity with another player will no longer be run.
# Objective
- Fixes#5529
## Solution
- Add assosciated constants named DEFAULT to as many types as possible
- Add const to as many methods in bevy_ui as possible
I have not applied the same treatment to the bundles in bevy_ui as it would require going into other bevy crates to implement const defaults for structs in bevy_text or relies on UiImage which calls HandleUntyped.typed() which isn't const safe.
Alternatively the defaults could relatively easily be turned into a macro to regain some of the readability and conciseness at the cost of explicitness.
Such a macro that partially implements this exists as a crate here: [const-default](https://docs.rs/const-default/latest/const_default/derive.ConstDefault.html) but does not support enums.
Let me know if there's anything I've missed or if I should push further into other crates.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
Speed up the render phase of rendering. Simplify the trait structure for render commands.
## Solution
- Merge `EntityPhaseItem` into `PhaseItem` (`EntityPhaseItem::entity` -> `PhaseItem::entity`)
- Merge `EntityRenderCommand` into `RenderCommand`.
- Add two associated types to `RenderCommand`: `RenderCommand::ViewWorldQuery` and `RenderCommand::WorldQuery`.
- Use the new associated types to construct two `QueryStates`s for `RenderCommandState`.
- Hoist any `SQuery<T>` fetches in `EntityRenderCommand`s into the aformentioned two queries. Batch fetch them all at once.
## Performance
`main_opaque_pass_3d` is slightly faster on `many_foxes` (427.52us -> 401.15us)
The shadow pass node is also slightly faster (344.52 -> 338.24us)
## Future Work
- Can we hoist the view level queries out of the core loop?
---
## Changelog
Added: `PhaseItem::entity`
Added: `RenderCommand::ViewWorldQuery` associated type.
Added: `RenderCommand::ItemorldQuery` associated type.
Added: `Draw<T>::prepare` optional trait function.
Removed: `EntityPhaseItem` trait
## Migration Guide
TODO
# Objective
- The #7064 PR had poor performance on an M1 Max in MacOS due to significant overuse of registers resulting in 'register spilling' where data that would normally be stored in registers on the GPU is instead stored in VRAM. The latency to read from/write to VRAM instead of registers incurs a significant performance penalty.
- Use of registers is a limiting factor in shader performance. Assignment of a struct from memory to a local variable can incur copies. Passing a variable that has struct type as an argument to a function can also incur copies. As such, these two cases can incur increased register usage and decreased performance.
## Solution
- Remove/avoid a number of assignments of light struct type data to local variables.
- Remove/avoid a number of passing light struct type variables/data as value arguments to shader functions.
# Objective
- The recently merged PR #7013 does not allow multiple `RenderPhase`s to share the same `RenderPass`.
- Due to the introduced overhead we want to minimize the number of `RenderPass`es recorded during each frame.
## Solution
- Take a constructed `TrackedRenderPass` instead of a `RenderPassDiscriptor` as a parameter to the `RenderPhase::render` method.
---
## Changelog
To enable multiple `RenderPhases` to share the same `TrackedRenderPass`,
the `RenderPhase::render` signature has changed.
```rust
pub fn render<'w>(
&self,
render_pass: &mut TrackedRenderPass<'w>,
world: &'w World,
view: Entity)
```
Co-authored-by: Kurt Kühnert <51823519+kurtkuehnert@users.noreply.github.com>
# Objective
`Query::get` and other random access methods require looking up `EntityLocation` for every provided entity, then always looking up the `Archetype` to get the table ID and table row. This requires 4 total random fetches from memory: the `Entities` lookup, the `Archetype` lookup, the table row lookup, and the final fetch from table/sparse sets. If `EntityLocation` contains the table ID and table row, only the `Entities` lookup and the final storage fetch are required.
## Solution
Add `TableId` and table row to `EntityLocation`. Ensure it's updated whenever entities are moved around. To ensure `EntityMeta` does not grow bigger, both `TableId` and `ArchetypeId` have been shrunk to u32, and the archetype index and table row are stored as u32s instead of as usizes. This should shrink `EntityMeta` by 4 bytes, from 24 to 20 bytes, as there is no padding anymore due to the change in alignment.
This idea was partially concocted by @BoxyUwU.
## Performance
This should restore the `Query::get` "gains" lost to #6625 that were introduced in #4800 without being unsound, and also incorporates some of the memory usage reductions seen in #3678.
This also removes the same lookups during add/remove/spawn commands, so there may be a bit of a speedup in commands and `Entity{Ref,Mut}`.
---
## Changelog
Added: `EntityLocation::table_id`
Added: `EntityLocation::table_row`.
Changed: `World`s can now only hold a maximum of 2<sup>32</sup>- 1 archetypes.
Changed: `World`s can now only hold a maximum of 2<sup>32</sup> - 1 tables.
## Migration Guide
A `World` can only hold a maximum of 2<sup>32</sup> - 1 archetypes and tables now. If your use case requires more than this, please file an issue explaining your use case.
# Objective
There isn't really a way to test that code using bevy_reflect compiles or doesn't compile for certain scenarios. This would be especially useful for macro-centric PRs like #6511 and #6042.
## Solution
Using `bevy_ecs_compile_fail_tests` as reference, added the `bevy_reflect_compile_fail_tests` crate.
Currently, this crate contains a very simple test case. This is so that we can get the basic foundation of this crate agreed upon and merged so that more tests can be added by other PRs.
### Open Questions
- [x] Should this be added to CI? (Answer: Yes)
---
## Changelog
- Added the `bevy_reflect_compile_fail_tests` crate for testing compilation errors
# Objective
Solve #5464
## Solution
Adds a `SystemInformationDiagnosticsPlugin` to add diagnostics.
Adds `Cargo.toml` flags to fix building on different platforms.
---
## Changelog
Adds `sysinfo` crate to `bevy-diagnostics`.
Changes in import order are due to clippy.
Co-authored-by: l1npengtul <35755164+l1npengtul@users.noreply.github.com>
Co-authored-by: IceSentry <c.giguere42@gmail.com>
# Objective
`TEXTURE_ADAPTER_SPECIFIC_FORMAT_FEATURES` was already included in `adapter.features()` on non-wasm target, and since it is the default value for `WgpuSettings.features`, the subsequent code will also combine into this feature:
b6066c30b6/crates/bevy_render/src/renderer/mod.rs (L155-L156)
# Objective
Bevy uses custom `Ptr` types so the rust borrow checker can help ensure lifetimes are correct, even when types aren't known. However, these types don't benefit from the automatic lifetime coercion regular rust references enjoy
## Solution
Add a couple methods to Ptr, PtrMut, and MutUntyped to allow for easy usage of these types in more complex scenarios.
## Changelog
- Added `as_mut` and `as_ref` methods to `MutUntyped`.
- Added `shrink` and `as_ref` methods to `PtrMut`.
## Migration Guide
- `MutUntyped::into_inner` now marks things as changed.
# Objective
All `RenderPhases` follow the same render procedure.
The same code is duplicated multiple times across the codebase.
## Solution
I simply extracted this code into a method on the `RenderPhase`.
This avoids code duplication and makes setting up new `RenderPhases` easier.
---
## Changelog
### Changed
You can now set up the rendering code of a `RenderPhase` directly using the `RenderPhase::render` method, instead of implementing it manually in your render graph node.
# Objective
- Loading a gltf files prints many errors
```
ERROR bevy_ecs::world: Unable to send event `bevy_hierarchy::events::HierarchyEvent`
Event must be added to the app with `add_event()`
https://docs.rs/bevy/*/bevy/app/struct.App.html#method.add_event
```
- Loading a gltf file create a world for a scene where events are not registered. Executing hierarchy commands on that world should not print error
## Solution
- Revert part of #6921
- don't use `world.send_event` / `world.send_event_batch` from commands
# Objective
- We already log the adapter info on startup when bevy_render is present. It would be nice to have more info about the system to be able to ask users to submit it in bug reports
## Solution
- Use the `sysinfo` crate to get all the information
- I made sure it _only_ gets the required informations to avoid unnecessary system request
- Add a system that logs this on startup
- This system is currently in `bevy_diagnostics` because I didn't really know where to put it.
Here's an example log from my system:
```log
INFO bevy_diagnostic: SystemInformation { os: "Windows 10 Pro", kernel: "19044", cpu: "AMD Ryzen 7 5800X 8-Core Processor", core_count: "8", memory: "34282242 KB" }
```
---
## Changelog
- Added a new default log when starting a bevy app that logs the system information
# Objective
It is often necessary to update an entity's parent while keeping its GlobalTransform static. Currently it is cumbersome and error-prone (two questions in the discord `#help` channel in the past week)
- Part 1 of #5475
- Part 2: #7024.
## Solution
- Add a `reparented_to` method to `GlobalTransform`
---
## Changelog
- Add a `reparented_to` method to `GlobalTransform`
# Objective
Resolve#6156.
The most common type of command is one that runs for a single entity. Built-in commands like this can be ergonomically added to the command queue using the `EntityCommands` struct. However, adding custom entity commands to the queue is quite cumbersome. You must first spawn an entity, store its ID in a local, then construct a command using that ID and add it to the queue. This prevents method chaining, which is the main benefit of using `EntityCommands`.
### Example (before)
```rust
struct MyCustomCommand(Entity);
impl Command for MyCustomCommand { ... }
let id = commands.spawn((...)).id();
commmands.add(MyCustomCommand(id));
```
## Solution
Add the `EntityCommand` trait, which allows directly adding per-entity commands to the `EntityCommands` struct.
### Example (after)
```rust
struct MyCustomCommand;
impl EntityCommand for MyCustomCommand { ... }
commands.spawn((...)).add(MyCustomCommand);
```
---
## Changelog
- Added the trait `EntityCommand`. This is a counterpart of `Command` for types that execute code for a single entity.
## Future Work
If we feel its necessary, we can simplify built-in commands (such as `Despawn`) to use this trait.
# Objective
Any closure with the signature `FnOnce(&mut World)` implicitly implements the trait `Command` due to a blanket implementation. However, this implementation unnecessarily has the `Sync` bound, which limits the types that can be used.
## Solution
Remove the bound.
---
## Changelog
- `Command` closures no longer need to implement the marker trait `std::marker::Sync`.
# Objective
The documentation for camera priority is very confusing at the moment, it requires a bit of "double negative" kind of thinking.
# Solution
Flipping the wording on the documentation to reflect more common usecases like having an overlay camera and also renaming it to "order", since priority implies that it will override the other camera rather than have both run.
Consolidation of all the feedback about #6271 as well as the addition of an "unconditionally visible" mode.
# Objective
The current implementation of the `Visibility` struct simply wraps a boolean.. which seems like an odd pattern when rust has such nice enums that allow for more expression using pattern-matching.
Additionally as it stands Bevy only has two settings for visibility of an entity:
- "unconditionally hidden" `Visibility { is_visible: false }`,
- "inherit visibility from parent" `Visibility { is_visible: true }`
where a root level entity set to "inherit" is visible.
Note that given the behaviour, the current naming of the inner field is a little deceptive or unclear.
Using an enum for `Visibility` opens the door for adding an extra behaviour mode. This PR adds a new "unconditionally visible" mode, which causes an entity to be visible even if its Parent entity is hidden. There should not really be any performance cost to the addition of this new mode.
--
The recently added `toggle` method is removed in this PR, as its semantics could be confusing with 3 variants.
## Solution
Change the Visibility component into
```rust
enum Visibility {
Hidden, // unconditionally hidden
Visible, // unconditionally visible
Inherited, // inherit visibility from parent
}
```
---
## Changelog
### Changed
`Visibility` is now an enum
## Migration Guide
- evaluation of the `visibility.is_visible` field should now check for `visibility == Visibility::Inherited`.
- setting the `visibility.is_visible` field should now directly set the value: `*visibility = Visibility::Inherited`.
- usage of `Visibility::VISIBLE` or `Visibility::INVISIBLE` should now use `Visibility::Inherited` or `Visibility::Hidden` respectively.
- `ComputedVisibility::INVISIBLE` and `SpatialBundle::VISIBLE_IDENTITY` have been renamed to `ComputedVisibility::HIDDEN` and `SpatialBundle::INHERITED_IDENTITY` respectively.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
- Be able to name the type that `ManualEventReader::iter/iter_with_id` returns and `EventReader::iter/iter_with_id` by proxy.
Currently for the purpose of https://github.com/bevyengine/bevy/pull/5719
## Solution
- Create a custom `Iterator` type.
# Objective
- alternative to #2895
- as mentioned in #2535 the uuid based ids in the render module should be replaced with atomic-counted ones
## Solution
- instead of generating a random UUID for each render resource, this implementation increases an atomic counter
- this might be replaced by the ids of wgpu if they expose them directly in the future
- I have not benchmarked this solution yet, but this should be slightly faster in theory.
- Bevymark does not seem to be affected much by this change, which is to be expected.
- Nothing of our API has changed, other than that the IDs have lost their IMO rather insignificant documentation.
- Maybe the documentation could be added back into the macro, but this would complicate the code.
# Objective
- Describe the objective or issue this PR addresses.
SpritePipelineKey could use more constification.
## Solution
Constify SpritePipelineKey implementation.
## Changelog
Co-authored-by: AxiomaticSemantics <117950168+AxiomaticSemantics@users.noreply.github.com>
# Objective
Align the hierarchy API between `EntityCommands` and `EntityMut`.
Added missing methods to `EntityMut`.
Replaced the duplicate `Command` implementations with the ones on `EntityMut` (e.g. The `AddChild` command is now just `world.entity_mut(..).add_child(..)`)
Fixed `update_old_parents` not sending `ChildAdded` events.
This PR does not add `add_children` to `EntityMut` as I would like to remove it from `EntityCommands` instead in #6942.
## Changelog
* Added `add_child`, `set_parent` and `remove_parent` to `EntityMut`
* Fixed missing `ChildAdded` events
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
* Currently, the `SystemParam` derive does not support types with const generic parameters.
* If you try to use const generics, the error message is cryptic and unhelpful.
* Continuation of the work started in #6867 and #6957.
## Solution
Allow const generic parameters to be used with `#[derive(SystemParam)]`.
# Objective
Fixes#4729.
Continuation of #4854.
## Solution
Add documentation to `ParamSet` and its methods. Includes examples suggested by community members in the original PR.
Co-authored-by: Nanox19435 <50684926+Nanox19435@users.noreply.github.com>
Co-authored-by: JoJoJet <21144246+JoJoJet@users.noreply.github.com>
# Objective
Fixes#4231.
## Solution
This PR implements the solution suggested by @bjorn3 : Use an internal property within `App` to detect `App::run()` calls from `Plugin::build()`.
---
## Changelog
- panic when App::run() is called from Plugin::build()
# Objective
Upgrade to Taffy 0.2
## Solution
Do it
## Changelog
Upgraded to Taffy 0.2, improving UI layout performance significantly and adding the flexbox `gap` property and `AlignContent::SpaceEvenly`.
## Notes
`many_buttons` is 8% faster! speed improvements for more highly nested UIs will be much more dramatic. Great work, Team Taffy.
# Objective
* The `SystemParam` derive internally uses tuples, which means it is constrained by the 16-field limit on `all_tuples`.
* The error message if you exceed this limit is abysmal.
* Supercedes #5965 -- this does the same thing, but is simpler.
## Solution
If any tuples have more than 16 fields, they are folded into tuples of tuples until they are under the 16-field limit.
# Objective
Currently, only named structs can be used with the `SystemParam` derive macro.
## Solution
Remove the restriction. Tuple structs and unit structs are now supported.
---
## Changelog
+ Added support for tuple structs and unit structs to the `SystemParam` derive macro.
# Objective
Fixes#6862 (oh hey good catch @alice-i-cecile)
Bevy was failing to print events from `info!()` and friends to the console if the `trace_tracy` feature was enabled. It shouldn't be doing that.
## Solution
The problem was this per-layer filter that was added in #4320 to suppress a noisy per-frame event (which Tracy requires in order to properly close out a frame):
- The problem event's target was `"bevy_render::renderer"`, not `"tracy"`. - So, the filter wasn't specifically targeting the noisy event.
- Without a default, `tracing_subscriber::filter::Targets` will remove _everything_ that doesn't match an explicit target rule. - So, the filter _was_ silencing the noisy event, along with everything else.
This PR changes that filter to do what was probably intended in #4320: suppress ~any events more verbose than `ERROR` from `bevy_render::renderer`~ the one problematically noisy event, but allow anything else that already made it through the top-level filter_layer.
Also, adds a comment to clarify the intent of that filter, since it's otherwise a bit opaque and required some research.
---
## Changelog
Fixed a bug that hid console log messages when the `trace_tracy` feature was enabled.
The Camera link in the UiCameraConfig was not rendered properly by the documentation.
# Objective
- In the UiCameraConfig page (https://docs.rs/bevy/latest/bevy/prelude/struct.UiCameraConfig.html), a link to the Camera page (https://docs.rs/bevy/latest/bevy/render/camera/struct.Camera.html) is broken.
## Solution
- It seems that when using URL fragment specifiers, backtick should not be used. It might be an issue of rust itself. Replacing the URL fragment specifier `[`Camera`]: bevy_render:📷:Camera` with `[Camera]: bevy_render:📷:Camera` solves this.
# Objective
- `bevy_ptr::{Ptr, PtrMut, OwnedPtr}` wrap raw pointers and should be printable using pointer formatting.
## Solution
- Add a `core::fmt::Pointer` impl for `Ptr`, `PtrMut` and `OwnedPtr` based on the wrapped `NonNull` pointer.
---
## Changelog
- Added a `core::fmt::Pointer` impl to `Ptr`, `PtrMut` and `OwnedPtr`.
Co-authored-by: MrGunflame <mrgunflame@protonmail.com>
## Objective
Bevy UI uses a `MeasureFunc` that preserves the aspect ratio of text, not just images. This means that the extent of flex-items containing text may be calculated incorrectly depending on the ratio of the text size compared to the size of its containing node.
Fixes#6748
Related to #6724
with Bevy 0.9:
with this PR (accurately matching the behavior of Flexbox):
## Solution
Only perform the aspect ratio calculations if the uinode contains an image.
## Changelog
* Added a field `preserve_aspect_ratio` to `CalculatedSize`
* The `MeasureFunc` only preserves the aspect ratio when `preserve_aspect_ratio` is true.
* `update_image_calculated_size_system` sets `preserve_aspect_ratio` to true for nodes with images.
# Objective
The `WgpuSettings` resource is only used during plugin build. Move it into the `RenderPlugin` struct.
Changing these settings requires re-initializing the render context, which is currently not supported.
If it is supported in the future it should probably be more explicit than changing a field on a resource, maybe something similar to the `CreateWindow` event.
## Migration Guide
```rust
// Before (0.9)
App::new()
.insert_resource(WgpuSettings { .. })
.add_plugins(DefaultPlugins)
// After (0.10)
App::new()
.add_plugins(DefaultPlugins.set(RenderPlugin {
wgpu_settings: WgpuSettings { .. },
}))
```
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
Following #4402, extract systems run on the render world instead of the main world, and allow retained state operations on it's resources. We're currently extracting to `ExtractedJoints` and then copying it twice during Prepare. Once into `SkinnedMeshJoints` and again into the actual GPU buffer.
This makes #4902 obsolete.
## Solution
Cut out the middle copy and directly extract joints into `SkinnedMeshJoints` and remove `ExtractedJoints` entirely.
This also removes the per-frame allocation that is being made to send `ExtractedJoints` into the render world.
## Performance
On my local machine, this halves the time for `prepare_skinned _meshes` on `many_foxes` (195.75us -> 93.93us on average).
---
## Changelog
Added: `BufferVec::truncate`
Added: `BufferVec::extend`
Changed: `SkinnedMeshJoints::build` now takes a `&mut BufferVec` instead of a `&mut Vec` as a parameter.
Removed: `ExtractedJoints`.
## Migration Guide
`ExtractedJoints` has been removed. Read the bound bones from `SkinnedMeshJoints` instead.
# Objective
Fix#1991. Allow users to have a bit more control over the creation and finalization of the threads in `TaskPool`.
## Solution
Add new methods to `TaskPoolBuilder` that expose callbacks that are called to initialize and finalize each thread in the `TaskPool`.
Unlike the proposed solution in #1991, the callback is argument-less. If an an identifier is needed, `std:🧵:current` should provide that information easily.
Added a unit test to ensure that they're being called correctly.
# Objective
- Dynamic scene builder can build scenes without components, if they didn't have any matching the type registry
- Those entities are not really useful in the final `DynamicScene`
## Solution
- Add a method `remove_empty_entities` that will remove empty entities. It's not called by default when calling `build`, I'm not sure if that's a good idea or not.
# Objective
There is currently no way to iterate over key/value pairs inside an `EntityMap`, which makes the usage of this struct very awkward. I couldn't think of a good reason why the `iter()` function should not be exposed, considering the interface already exposes `keys()` and `values()`, so I made this PR.
## Solution
Implement `iter()` for `EntityMap` in terms of its inner map type.
# Objective
Some settings were only applied in windowed mode.
Fix the issue in #6933
# Solution
Always apply the settings.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
Remove a method with an unfortunate name and questionable usefulness.
Added in #4708
It doesn't make sense to me for us to provide a method to work around a limitation of closures when we can simply, *not* use a closure.
The limitation in this case is not being able to initialize a variable from inside a closure:
```rust
let child_id;
commands.spawn_empty().with_children(|parent| {
// Error: passing uninitalized variable to a closure.
child_id = parent.spawn_empty().id();
});
// Do something with child_id
```
The docs for `add_children` suggest the following:
```rust
let child_id = commands
.spawn_empty()
.add_children(|parent| parent.spawn_empty().id());
```
I would instead suggest using the following snippet.
```rust
let parent_id = commands.spawn_empty().id();
let child_id = commands.spawn_empty().set_parent(parent_id).id();
// To be fair, at the time of #4708 this would have been a bit more cumbersome since `set_parent` did not exist.
```
Using `add_children` gets more unwieldy when you also want the `parent_id`.
```rust
let parent_commands = commands.spawn_empty();
let parent_id = parent_commands.id();
let child_id = parent_commands.add_children(|parent| parent.spawn_empty().id());
```
### The name
I see why `add_children` is named that way, it's the non-builder variant of `with_children` so it kinda makes sense,
but now the method name situation for `add_child`, `add_children` and `push_children` is *rather* unfortunate.
Removing `add_children` and renaming `push_children` to `add_children` in one go is kinda bleh, but that way we end up with the matching methods `add_child` and `add_children`.
Another reason to rename `push_children` is that it's trying to mimick the `Vec` api naming but fails because `push` is for single elements. I guess it should have been `extend_children_from_slice`, but lets not name it that :)
### Questions
~~Should `push_children` be renamed in this PR? This would make the migration guide easier to deal with.~~
Let's do that later.
Does anyone know of a way to do a simple text/regex search through all the github repos for usage of `add_children`?
That way we can have a better idea of how this will affect users. My guess is that usage of `add_children` is quite rare.
## Migration Guide
The method `add_children` on `EntityCommands` was removed.
If you were using `add_children` over `with_children` to return data out of the closure you can use `set_parent` or `add_child` to avoid the closure instead.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
`AsBindGroup` can't be used as a trait object because of the constraint `Sized` and because of the associated function.
This is a problem for [`bevy_atmosphere`](https://github.com/JonahPlusPlus/bevy_atmosphere) because it needs to use a trait that depends on `AsBindGroup` as a trait object, for switching out different shaders at runtime. The current solution it employs is reimplementing the trait and derive macro into that trait, instead of constraining to `AsBindGroup`.
## Solution
Remove the `Sized` constraint from `AsBindGroup` and add the constraint `where Self: Sized` to the associated function `bind_group_layout`. Also change `PreparedBindGroup<T: AsBindGroup>` to `PreparedBindGroup<T>` and use it as `PreparedBindGroup<Self::Data>` instead of `PreparedBindGroup<Self>`.
This weakens the constraints, but increases the flexibility of `AsBindGroup`.
I'm not entirely sure why the `Sized` constraint was there, because it worked fine without it (maybe @cart wasn't aware of use cases for `AsBindGroup` as a trait object or this was just leftover from legacy code?).
---
## Changelog
- `AsBindGroup` can be used as a trait object.
# Objective
[Rust 1.66](https://blog.rust-lang.org/inside-rust/2022/12/12/1.66.0-prerelease.html) is coming in a few days, and bevy doesn't build with it.
Fix that.
## Solution
Replace output from a trybuild test, and fix a few new instances of `needless_borrow` and `unnecessary_cast` that are now caught.
## Note
Due to the trybuild test, this can't be merged until 1.66 is released.
# Objective
A separate `tracing` span for running a system's commands is created, even if the system doesn't have commands. This is adding extra measuring overhead (see #4892) where it's not needed.
## Solution
Move the span into `ParallelCommandState` and `CommandQueue`'s `SystemParamState::apply`. To get the right metadata for the span, a additional `&SystemMeta` parameter was added to `SystemParamState::apply`.
---
## Changelog
Added: `SystemMeta::name`
Changed: Systems without `Commands` and `ParallelCommands` will no longer show a "system_commands" span when profiling.
Changed: `SystemParamState::apply` now takes a `&SystemMeta` parameter in addition to the provided `&mut World`.
# Objective
Change detection can be spuriously triggered by setting a field to the same value as before. As a result, a common pattern is to write:
```rust
if *foo != value {
*foo = value;
}
```
This is confusing to read, and heavy on boilerplate.
Adopted from #5373, but untangled and rebased to current `bevy/main`.
## Solution
1. Add a method to the `DetectChanges` trait that implements this boilerplate when the appropriate trait bounds are met.
2. Document this minor footgun, and point users to it.
## Changelog
* added the `set_if_neq` method to avoid triggering change detection when the new and previous values are equal. This will work on both components and resources.
## Migration Guide
If you are manually checking if a component or resource's value is equal to its new value before setting it to avoid triggering change detection, migrate to the clearer and more convenient `set_if_neq` method.
## Context
Related to #2363 as it avoids triggering change detection, but not a complete solution (as it still requires triggering it when real changes are made).
Co-authored-by: Zoey <Dessix@Dessix.net>
# Objective
The following code:
```rs
use bevy::prelude::Image;
use image::{ DynamicImage, GenericImage, Rgba };
fn main() {
let mut dynamic_image = DynamicImage::new_rgb32f(1, 1);
dynamic_image.put_pixel(0, 0, Rgba([1, 1, 1, 1]));
let image = Image::from_dynamic(dynamic_image, false); // Panic!
println!("{image:?}");
}
```
Can cause an assertion failed:
```
thread 'main' panicked at 'assertion failed: `(left == right)`
left: `16`,
right: `14`: Pixel data, size and format have to match', .../bevy_render-0.9.1/src/texture/image.rs:209:9
stack backtrace:
...
4: core::panicking::assert_failed<usize,usize>
at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:181
5: bevy_render::texture::image::Image::new
at .../bevy_render-0.9.1/src/texture/image.rs:209
6: bevy_render::texture::image::Image::from_dynamic
at .../bevy_render-0.9.1/src/texture/image_texture_conversion.rs:159
7: bevy_test::main
at ./src/main.rs:8
...
```
It seems to be cause by a copypasta in `crates/bevy_render/src/texture/image_texture_conversion.rs`. Let's fix it.
## Solution
```diff
// DynamicImage::ImageRgb32F(image) => {
- let a = u16::max_value();
+ let a = 1f32;
```
This will fix the conversion.
---
## Changelog
- Fixed the alpha channel of the `image::DynamicImage::ImageRgb32F` to `bevy_render::texture::Image` conversion in `bevy_render::texture::Image::from_dynamic()`.
# Objective
- https://github.com/bevyengine/bevy/pull/5364 Added a few features to the AsBindGroup derive, but if you don't know they exist they aren't documented anywhere.
## Solution
- Document the new arguments in the doc block for the derive.
# Objective
Speed up bundle insertion and spawning from a bundle.
## Solution
Use the same technique used in #6800 to remove the branch on storage type when writing components from a `Bundle` into storage.
- Add a `StorageType` argument to the closure on `Bundle::get_components`.
- Pass `C::Storage::STORAGE_TYPE` into that argument.
- Match on that argument instead of reading from a `Vec<StorageType>` in `BundleInfo`.
- Marked all implementations of `Bundle::get_components` as inline to encourage dead code elimination.
The `Vec<StorageType>` in `BundleInfo` was also removed as it's no longer needed. If users were reliant on this, they can either use the compile time constants or fetch the information from `Components`. Should save a rather negligible amount of memory.
## Performance
Microbenchmarks show a slight improvement to inserting components into existing entities, as well as spawning from a bundle. Ranging about 8-16% faster depending on the benchmark.
```
group main soft-constant-write-components
----- ---- ------------------------------
add_remove/sparse_set 1.08 1019.0±80.10µs ? ?/sec 1.00 944.6±66.86µs ? ?/sec
add_remove/table 1.07 1343.3±20.37µs ? ?/sec 1.00 1257.3±18.13µs ? ?/sec
add_remove_big/sparse_set 1.08 1132.4±263.10µs ? ?/sec 1.00 1050.8±240.74µs ? ?/sec
add_remove_big/table 1.02 2.6±0.05ms ? ?/sec 1.00 2.5±0.08ms ? ?/sec
get_or_spawn/batched 1.15 401.4±17.76µs ? ?/sec 1.00 349.3±11.26µs ? ?/sec
get_or_spawn/individual 1.13 732.1±43.35µs ? ?/sec 1.00 645.6±41.44µs ? ?/sec
insert_commands/insert 1.12 623.9±37.48µs ? ?/sec 1.00 557.4±34.99µs ? ?/sec
insert_commands/insert_batch 1.16 401.4±17.00µs ? ?/sec 1.00 347.4±12.87µs ? ?/sec
insert_simple/base 1.08 416.9±5.60µs ? ?/sec 1.00 385.2±4.14µs ? ?/sec
insert_simple/unbatched 1.06 934.5±44.58µs ? ?/sec 1.00 881.3±47.86µs ? ?/sec
spawn_commands/2000_entities 1.09 190.7±11.41µs ? ?/sec 1.00 174.7±9.15µs ? ?/sec
spawn_commands/4000_entities 1.10 386.5±25.33µs ? ?/sec 1.00 352.3±18.81µs ? ?/sec
spawn_commands/6000_entities 1.10 586.2±34.42µs ? ?/sec 1.00 535.3±27.25µs ? ?/sec
spawn_commands/8000_entities 1.08 778.5±45.15µs ? ?/sec 1.00 718.0±33.66µs ? ?/sec
spawn_world/10000_entities 1.04 1026.4±195.46µs ? ?/sec 1.00 985.8±253.37µs ? ?/sec
spawn_world/1000_entities 1.06 103.8±20.23µs ? ?/sec 1.00 97.6±18.22µs ? ?/sec
spawn_world/100_entities 1.15 11.4±4.25µs ? ?/sec 1.00 9.9±1.87µs ? ?/sec
spawn_world/10_entities 1.05 1030.8±229.78ns ? ?/sec 1.00 986.2±231.12ns ? ?/sec
spawn_world/1_entities 1.01 105.1±23.33ns ? ?/sec 1.00 104.6±31.84ns ? ?/sec
```
---
## Changelog
Changed: `Bundle::get_components` now takes a `FnMut(StorageType, OwningPtr)`. The provided storage type must be correct for the component being fetched.
# Objective
Partially address #3492.
## Solution
Document the remaining undocumented members of `bevy_utils` and set `warn(missing_docs)` on the crate level. Also enabled `clippy::undocumented_unsafe_blocks` as a warning on the crate to keep it in sync with `bevy_ecs`'s warnings.
# Objective
```rust
// makes clippy complain about 'taking a mutable reference to a `const` item'
let color = *Color::RED.set_a(0.5);
// Now you can do
let color = Color::RED.with_a(0.5);
```
## Changelog
Added `with_r`, `with_g`, `with_b`, and `with_a` to `Color`.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
- Fixes https://github.com/bevyengine/bevy/issues/6417
## Solution
- clear_trackers was not being called on the render world. This causes the removed components vecs to continuously grow. This PR adds clear trackers to the end of RenderStage::Cleanup
## Migration Guide
The call to `clear_trackers` in `App` has been moved from the schedule to App::update for the main world and calls to `clear_trackers` have been added for sub_apps in the same function. This was due to needing stronger guarantees. If clear_trackers isn't called on a world it can lead to memory leaks in `RemovedComponents`.
# Objective
* Implementing a custom `SystemParam` by hand requires implementing three traits -- four if it is read-only.
* The trait `SystemParamFetch<'w, 's>` is a workaround from before we had generic associated types, and is no longer necessary.
## Solution
* Combine the trait `SystemParamFetch` with `SystemParamState`.
* I decided to remove the `Fetch` name and keep the `State` name, since the former was consistently conflated with the latter.
* Replace the trait `ReadOnlySystemParamFetch` with `ReadOnlySystemParam`, which simplifies trait bounds in generic code.
---
## Changelog
- Removed the trait `SystemParamFetch`, moving its functionality to `SystemParamState`.
- Replaced the trait `ReadOnlySystemParamFetch` with `ReadOnlySystemParam`.
## Migration Guide
The trait `SystemParamFetch` has been removed, and its functionality has been transferred to `SystemParamState`.
```rust
// Before
impl SystemParamState for MyParamState {
fn init(world: &mut World, system_meta: &mut SystemMeta) -> Self { ... }
}
impl<'w, 's> SystemParamFetch<'w, 's> for MyParamState {
type Item = MyParam<'w, 's>;
fn get_param(...) -> Self::Item;
}
// After
impl SystemParamState for MyParamState {
type Item<'w, 's> = MyParam<'w, 's>; // Generic associated types!
fn init(world: &mut World, system_meta: &mut SystemMeta) -> Self { ... }
fn get_param<'w, 's>(...) -> Self::Item<'w, 's>;
}
```
The trait `ReadOnlySystemParamFetch` has been replaced with `ReadOnlySystemParam`.
```rust
// Before
unsafe impl ReadOnlySystemParamFetch for MyParamState {}
// After
unsafe impl<'w, 's> ReadOnlySystemParam for MyParam<'w, 's> {}
```
# Objective
It's not clear to users how to handle `!Sync` types as components and resources in the absence of engine level support for them.
## Solution
Added a section to `Component`'s and `Resource`'s type level docs on available options for making a type `Sync` when it holds `!Sync` fields, linking `bevy_utils::synccell::SyncCell` and the currently unstable `std::sync::Exclusive`.
Also added a compile_fail doctest that illustrates how to apply `SyncCell`. These will break when/if #6572 gets merged, at which point these docs should be updated.
# Objective
This is an adoption of #5792. Fixes#5791.
## Solution
Implemented all the required reflection traits for `VecDeque`, taking from `Vec`'s impls.
---
## Changelog
Added: `std::collections::VecDeque` now implements `Reflect` and all relevant traits.
Co-authored-by: james7132 <contact@jamessliu.com>
# Objective
- Get rid of giant match statement to get PixelInfo.
- This will allow for supporting any texture that is uncompressed, instead of people needing to PR in any textures that are supported in wgpu, but not bevy.
## Solution
- More conservative alternative to https://github.com/bevyengine/bevy/pull/6788, where we don't try to make some of the calculations correct for compressed types.
- Delete `PixelInfo` and get the pixel_size directly from wgpu. Data from wgpu is here: https://docs.rs/wgpu-types/0.14.0/src/wgpu_types/lib.rs.html#2359
- Panic if the texture is a compressed type. An integer byte size of a pixel is no longer a valid concept when talking about compressed textures.
- All internal usages use `pixel_size` and not `pixel_info` and are on uncompressed formats. Most of these usages are on either explicit texture formats or slightly indirectly through `TextureFormat::bevy_default()`. The other uses are in `TextureAtlas` and have other calculations that assumes the texture is uncompressed.
## Changelog
- remove `PixelInfo` and get `pixel_size` from wgpu
## Migration Guide
`PixelInfo` has been removed. `PixelInfo::components` is equivalent to `texture_format.describe().components`. `PixelInfo::type_size` can be gotten from `texture_format.describe().block_size/ texture_format.describe().components`. But note this can yield incorrect results for some texture types like Rg11b10Float.
# Objective
Adds a cylinder shape. Fixes#2282.
## Solution
- I added a custom cylinder shape, taken from [here](https://github.com/rparrett/typey_birb/blob/main/src/cylinder.rs) with permission from @rparrett.
- I also added the cylinder shape to the `3d_shapes` example scene.
---
## Changelog
- Added cylinder shape
Co-Authored-By: Rob Parrett <robparrett@gmail.com>
Co-Authored-By: davidhof <7483215+davidhof@users.noreply.github.com>
# Objective
Fixes#6224, add ``dbg``, ``info``, ``warn`` and ``error`` system piping adapter variants to expand #5776, which call the corresponding re-exported [bevy_log macros](https://docs.rs/bevy/latest/bevy/log/macro.info.html) when the result is an error.
## Solution
* Added ``dbg``, ``info``, ``warn`` and ``error`` system piping adapter variants to ``system_piping.rs``.
* Modified and added tests for these under examples in ``system_piping.rs``.
# Objective
- Make `AudioOutput` a `Resource`.
## Solution
- Do not store `OutputStream` in the struct.
- `mem::forget` `OutputStream`.
---
## Changelog
### Added
- `AudioOutput` is now a `Resource`.
## Migration Guide
- Use `Res<AudioOutput<Source>>` instead of `NonSend<AudioOutput<Source>>`. Same for `Mut` variants.
# Objective
Resolves#4597 (based on the work from #6056 and a refresh of #4147)
When using reflection, we may often end up in a scenario where we have a Dynamic representing a certain type. Unfortunately, we can't just call `MyType::from_reflect` as we do not have knowledge of the concrete type (`MyType`) at runtime.
Such scenarios happen when we call `Reflect::clone_value`, use the reflection deserializers, or create the Dynamic type ourselves.
## Solution
Add a `ReflectFromReflect` type data struct.
This struct allows us to easily convert Dynamic representations of our types into their respective concrete instances.
```rust
#[derive(Reflect, FromReflect)]
#[reflect(FromReflect)] // <- Register `ReflectFromReflect`
struct MyStruct(String);
let type_id = TypeId::of::<MyStruct>();
// Register our type
let mut registry = TypeRegistry::default();
registry.register::<MyStruct>();
// Create a concrete instance
let my_struct = MyStruct("Hello world".to_string());
// `Reflect::clone_value` will generate a `DynamicTupleStruct` for tuple struct types
let dynamic_value: Box<dyn Reflect> = my_struct.clone_value();
assert!(!dynamic_value.is::<MyStruct>());
// Get the `ReflectFromReflect` type data from the registry
let rfr: &ReflectFromReflect = registry
.get_type_data::<ReflectFromReflect>(type_id)
.unwrap();
// Call `FromReflect::from_reflect` on our Dynamic value
let concrete_value: Box<dyn Reflect> = rfr.from_reflect(&dynamic_value);
assert!(concrete_value.is::<MyStruct>());
```
### Why this PR?
###### Why now?
The three main reasons I closed#4147 were that:
1. Registering `ReflectFromReflect` is clunky (deriving `FromReflect` *and* registering `ReflectFromReflect`)
2. The ecosystem and Bevy itself didn't seem to pay much attention to deriving `FromReflect`
3. I didn't see a lot of desire from the community for such a feature
However, as time has passed it seems 2 and 3 are not really true anymore. Bevy is internally adding lots more `FromReflect` derives, which should make this feature all the more useful. Additionally, I have seen a growing number of people look for something like `ReflectFromReflect`.
I think 1 is still an issue, but not a horrible one. Plus it could be made much, much better using #6056. And I think splitting this feature out of #6056 could lead to #6056 being adopted sooner (or at least make the need more clear to users).
###### Why not just re-open #4147?
The main reason is so that this PR can garner more attention than simply re-opening the old one. This helps bring fresh eyes to the PR for potentially more perspectives/reviews.
---
## Changelog
* Added `ReflectFromReflect`
Co-authored-by: Gino Valente <49806985+MrGVSV@users.noreply.github.com>
# Objective
#6547 accidentally broke change detection for SparseSet components by using `Ticks::from_tick_cells` with the wrong argument order.
## Solution
Use the right argument order. Add a regression test.
# Objective
The window event types currently don't support reflection. This PR adds support to them (as requested [here](https://github.com/bevyengine/bevy/issues/6223#issuecomment-1273852329)).
## Solution
Implement `Reflect` + `FromReflect` for window event types. Relevant traits are also being reflected with `#[reflect(...)]` attributes.
Additionally, this PR derives `Reflect` + `FromReflect` for `WindowDescriptor` and the types it depends on so that `CreateWindow` events can be fully manipulated through reflection.
Finally, this PR adds `FromReflect` for `PathBuf` as a value type, which is needed for `FileDragAndDrop`.
This adds the "glam" feature to the `bevy_reflect` dependency for package `bevy_window`. Since `bevy_window` transitively depends on `glam` already, all this brings in are the reflection `impl`s.
## Open questions
Should `app.register_type::<PathBuf>();` be moved to `CorePlugin`? I added it to `WindowPlugin` because that's where it's used and `CorePlugin` doesn't seem to register all the missing std types, but it would also make sense in `CorePlugin` I believe since it's a commonly used type.
---
## Changelog
Added:
- Implemented `Reflect` + `FromReflect` for window events and related types. These types are automatically registered when adding the `WindowPlugin`.
# Objective
- Fixes#6841
- In some case, the number of maximum storage buffers is `u32::MAX` which doesn't fit in a `i32`
## Solution
- Add an option to have a `u32` in a `ShaderDefVal`
having `doc(hidden)` on the read only version of a generated mutable world query leads to docs on the readonly item having a dead link. It also makes it annoying to have nice docs for libraries attempting to expose derived `WorldQuery` structs as re-exporting the read only item does not cause it to appear in docs even though it would be intended for users to know about the read only world query and use it.
# Objective
Fixes#6866.
## Solution
Docs now should describe what the _front_, _first_, _back_, and _last_ elements are for an implementor of the `bevy::reflect::list::List` Trait. Further, the docs should describe how `bevy::reflect::list::List::push` and `bevy::reflect::list::List::pop` should act on these elements.
Co-authored-by: Linus Käll <linus.kall.business@gmail.com>
# Objective
Prevent future unsoundness that was seen in #6623.
## Solution
Newtype both indexes in `Archetype` and `Table` as `ArchetypeRow` and `TableRow`. This avoids weird numerical manipulation on the indices, and can be stored and treated opaquely. Also enforces the source and destination of where these indices at a type level.
---
## Changelog
Changed: `Archetype` indices and `Table` rows have been newtyped as `ArchetypeRow` and `TableRow`.
# Objective
`EntityRef::get` and friends all type erase calls to fetch the target components by using passing in the `TypeId` instead of using generics. This is forcing a lookup to `Components` to fetch the storage type. This adds an extra memory lookup and forces a runtime branch instead of allowing the compiler to optimize out the unused branch.
## Solution
Leverage `Component::Storage::STORAGE_TYPE` as a constant instead of fetching the metadata from `Components`.
## Performance
This has a near 2x speedup for all calls to `World::get`. Microbenchmark results from my local machine. `Query::get_component`, which uses `EntityRef::get` internally also show a slight speed up. This has closed the gap between `World::get` and `Query::get` for the same use case.
```
group entity-ref-generics main
----- ------------------- ----
query_get_component/50000_entities_sparse 1.00 890.6±40.42µs ? ?/sec 1.10 980.6±28.22µs ? ?/sec
query_get_component/50000_entities_table 1.00 968.5±73.73µs ? ?/sec 1.08 1048.8±31.76µs ? ?/sec
query_get_component_simple/system 1.00 703.2±4.37µs ? ?/sec 1.00 702.1±6.13µs ? ?/sec
query_get_component_simple/unchecked 1.02 855.8±8.98µs ? ?/sec 1.00 843.1±8.19µs ? ?/sec
world_get/50000_entities_sparse 1.00 202.3±3.15µs ? ?/sec 1.85 374.0±20.96µs ? ?/sec
world_get/50000_entities_table 1.00 193.0±1.78µs ? ?/sec 2.02 389.2±26.55µs ? ?/sec
world_query_get/50000_entities_sparse 1.01 162.4±2.23µs ? ?/sec 1.00 161.3±0.95µs ? ?/sec
world_query_get/50000_entities_table 1.00 199.9±0.63µs ? ?/sec 1.00 200.2±0.74µs ? ?/sec
```
This should also, by proxy, speed up the `ReflectComponent` APIs as most of those use `World::get` variants internally.
# Objective
- Fixes#3004
## Solution
- Replaced all the types with their fully quallified names
- Replaced all trait methods and inherent methods on dyn traits with their fully qualified names
- Made a new file `fq_std.rs` that contains structs corresponding to commonly used Structs and Traits from `std`. These structs are replaced by their respective fully qualified names when used inside `quote!`
# Objective
- Fixes#6711
## Solution
- Change the `path` function parameter of `dynamically_load_plugin` and `DynamicPluginExt::load_plugin` to a generic with `AsRef<OsStr>` bound
# Objective
`prepare_asset` for Image has an alternate path for texture creation that is used when the image is not compressed and does not contain mipmaps. This additional code path is unnecessary as `render_device.create_texture_with_data()` will handle both cases correctly.
## Solution
Use `render_device.create_texture_with_data()` in all cases.
Tested successfully with the following examples:
- load_gltf
- render_to_texture
- texture
- 3d_shapes
- sprite
- sprite_sheet
- array_texture
- shader_material_screenspace_texture
- skybox (though this already would use the `create_texture_with_data()` branch anyway)
# Objective
The methods `World::change_tick` and `World::read_change_tick` lack documentation and have confusingly similar behavior.
## Solution
Add documentation and clarify the distinction between the two functions.
The PR fixes the interface of `EventReader::clear`. Currently, the method consumes the reader, which makes it unusable.
## Changelog
- `EventReader::clear` now takes a mutable reference instead of consuming the event reader.
## Migration Guide
`EventReader::clear` now takes a mutable reference instead of consuming the event reader. This means that `clear` now needs explicit mutable access to the reader variable, which previously could have been omitted in some cases:
```rust
// Old (0.9)
fn clear_events(reader: EventReader<SomeEvent>) {
reader.clear();
}
// New (0.10)
fn clear_events(mut reader: EventReader<SomeEvent>) {
reader.clear();
}
```
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
- Fixes#6812.
## Solution
- Replaced `World::read_change_ticks` with `World::change_ticks` within `bevy_ecs` crate in places where `World` references were mutable.
---
# Objective
Partially addresses #5504. Allow users to get an `Iterator<Item = EntityRef<'a>>` over all entities in the `World`.
## Solution
Change `World::iter_entities` to return an iterator of `EntityRef` instead of `Entity`.
Not sure how to tackle making an `Iterator<Item = EntityMut<'_>>` without being horribly unsound. Might need to wait for `LendingIterator` to stabilize so we can ensure only one of them is valid at a given time.
---
## Changelog
Changed: `World::iter_entities` now returns an iterator of `EntityRef` instead of `Entity`.
# Objective
Currently, the `SystemParam` derive forces you to declare the lifetime parameters `<'w, 's>`, even if you don't use them.
If you don't follow this structure, the error message is quite nasty.
### Example (before):
```rust
#[derive(SystemParam)]
pub struct EventWriter<'w, 's, E: Event> {
events: ResMut<'w, Events<E>>,
// The derive forces us to declare the `'s` lifetime even though we don't use it,
// so we have to add this `PhantomData` to please rustc.
#[system_param(ignore)]
_marker: PhantomData<&'s ()>,
}
```
## Solution
* Allow the user to omit either lifetime.
* Emit a descriptive error if any lifetimes used are invalid.
### Example (after):
```rust
#[derive(SystemParam)]
pub struct EventWriter<'w, E: Event> {
events: ResMut<'w, Events<E>>,
}
```
---
## Changelog
* The `SystemParam` derive is now more flexible, allowing you to omit unused lifetime parameters.
Updates the requirements on [tracing-chrome](https://github.com/thoren-d/tracing-chrome) to permit the latest version.
<details>
<summary>Release notes</summary>
<p><em>Sourced from <a href="https://github.com/thoren-d/tracing-chrome/releases">tracing-chrome's releases</a>.</em></p>
<blockquote>
<h2>Release v0.7.0</h2>
<ul>
<li>Add <code>start_new</code> to <code>FlushGuard</code>. You can now generate multiple traces in one run!</li>
<li>Clean up dependencies</li>
<li>Make events thread-scoped</li>
</ul>
</blockquote>
</details>
<details>
<summary>Commits</summary>
<ul>
<li><a href="8760d81206"><code>8760d81</code></a> Prepare for 0.7 (<a href="https://github-redirect.dependabot.com/thoren-d/tracing-chrome/issues/16">#16</a>)</li>
<li><a href="a30bfb78d2"><code>a30bfb7</code></a> Update documentation (<a href="https://github-redirect.dependabot.com/thoren-d/tracing-chrome/issues/15">#15</a>)</li>
<li><a href="3fe24ff44d"><code>3fe24ff</code></a> Save thread names for start_new (<a href="https://github-redirect.dependabot.com/thoren-d/tracing-chrome/issues/14">#14</a>)</li>
<li><a href="fa8a0ff8ba"><code>fa8a0ff</code></a> Adding "Start New" feature that allows a user to finalize writing to the (<a href="https://github-redirect.dependabot.com/thoren-d/tracing-chrome/issues/11">#11</a>)</li>
<li><a href="d3059d66b3"><code>d3059d6</code></a> Directly depend on crossbeam_channel (<a href="https://github-redirect.dependabot.com/thoren-d/tracing-chrome/issues/12">#12</a>)</li>
<li><a href="441dba5c21"><code>441dba5</code></a> change scope of instant events to thread (<a href="https://github-redirect.dependabot.com/thoren-d/tracing-chrome/issues/13">#13</a>)</li>
<li>See full diff in <a href="https://github.com/thoren-d/tracing-chrome/compare/v0.6.0...v0.7.0">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
Fixes#6827
## Solution
Use the `Color::rgba_linear` function instead of the `Color::rgba` function to correctly interpret colors from glTF files in the linear color space rather than the incorrect sRGB color space
# Objective
> Followup to [this](https://github.com/bevyengine/bevy/pull/6755#discussion_r1032671178) comment
Rearrange the impls in the `impls/std.rs` file.
The issue was that I had accidentally misplaced the impl for `Option<T>` and put it between the `Cow<'static, str>` impls. This is just a slight annoyance and readability issue.
## Solution
Move the `Option<T>` and `&'static Path` impls around to be more readable.
# Objective
The soundness of the ECS `World` partially relies on the correctness of the state of `Entities` stored within it. We're currently allowing users to (unsafely) mutate it, as well as readily construct it without using a `World`. While this is not strictly unsound so long as users (including `bevy_render`) safely use the APIs, it's a fairly easy path to unsoundness without much of a guard rail.
Addresses #3362 for `bevy_ecs::entity`. Incorporates the changes from #3985.
## Solution
Remove `Entities`'s `Default` implementation and force access to the type to only be through a properly constructed `World`.
Additional cleanup for other parts of `bevy_ecs::entity`:
- `Entity::index` and `Entity::generation` are no longer `pub(crate)`, opting to force the rest of bevy_ecs to use the public interface to access these values.
- `EntityMeta` is no longer `pub` and also not `pub(crate)` to attempt to cut down on updating `generation` without going through an `Entities` API. It's currently inaccessible except via the `pub(crate)` Vec on `Entities`, there was no way for an outside user to use it.
- Added `Entities::set`, an unsafe `pub(crate)` API for setting the location of an Entity (parallel to `Entities::get`) that replaces the internal case where we need to set the location of an entity when it's been spawned, moved, or despawned.
- `Entities::alloc_at_without_replacement` is only used in `World::get_or_spawn` within the first party crates, and I cannot find a public use of this API in any ecosystem crate that I've checked (via GitHub search).
- Attempted to document the few remaining undocumented public APIs in the module.
---
## Changelog
Removed: `Entities`'s `Default` implementation.
Removed: `EntityMeta`
Removed: `Entities::alloc_at_without_replacement` and `AllocAtWithoutReplacement`.
Co-authored-by: james7132 <contact@jamessliu.com>
Co-authored-by: James Liu <contact@jamessliu.com>
# Objective
- Since #5900 3d examples fail in wasm
```
ERROR crates/bevy_render/src/render_resource/pipeline_cache.rs:660 failed to process shader: Unknown shader def: 'AVAILABLE_STORAGE_BUFFER_BINDINGS'
```
## Solution
- Fix it by always adding the shaderdef `AVAILABLE_STORAGE_BUFFER_BINDINGS` with the actual value, instead of 3 when 3 or more were available
# Objective
- Support textures in `Rgb9e5Ufloat` format.
## Solution
- Add `TextureFormatPixelInfo` for `Rgb9e5Ufloat`.
Tested this with a `Rgb9e5Ufloat` encoded KTX2 texture.
# Objective
- Every usage of `DrawFunctionsInternals::get_id()` was followed by a `.unwrap()`. which just adds boilerplate.
## Solution
- Introduce a fallible version of `DrawFunctionsInternals::get_id()` and use it where possible.
- I also took the opportunity to improve the error message a little in the case where it fails.
---
## Changelog
- Added `DrawFunctionsInternals::id()`
# Objective
Document `bevy_ecs::archetype` and and declutter the public documentation for the module by making types non-`pub`.
Addresses #3362 for `bevy_ecs::archetype`.
## Solution
- Add module level documentation.
- Add type and API level documentation for all public facing types.
- Make `ArchetypeId`, `ArchetypeGeneration`, and `ArchetypeComponentId` truly opaque IDs that are not publicly constructable.
- Make `AddBundle` non-pub, make `Edges::get_add_bundle` return a `Option<ArchetypeId>` and fork the existing function into `Edges::get_add_bundle_internal`.
- Remove `pub(crate)` on fields that have a corresponding pub accessor function.
- Removed the `Archetypes: Default` impl, opting for a `pub(crate) fn new` alternative instead.
---
## Changelog
Added: `ArchetypeGeneration` now implements `Ord` and `PartialOrd`.
Removed: `Archetypes`'s `Default` implementation.
Removed: `Archetype::new` and `Archetype::is_empty`.
Removed: `ArchetypeId::new` and `ArchetypeId::value`.
Removed: `ArchetypeGeneration::value`
Removed: `ArchetypeIdentity`.
Removed: `ArchetypeComponentId::new` and `ArchetypeComponentId::value`.
Removed: `AddBundle`. `Edges::get_add_bundle` now returns `Option<ArchetypeId>`
# Objective
- The documentation describing different ways to spawn an Entity is missing reference to "method" for "Spawn an entity with components".
## Solution
- Update the documentation to add the reference to `World::spawn`.
# Objective
* Fix#6668
* There is no need to panic when a parenting operation is redundant, as no invalid state is entered.
## Solution
Make `push_children` idempotent.
# Objective
One of the use-cases for the function `Entity::from_raw` is creating placeholder entity ids, which are meant to be overwritten later. If we use a constant for this instead of `from_raw`, it is more ergonomic and self-documenting.
## Solution
Add a constant that returns an entity ID with an index of `u32::MAX` and a generation of zero. Users are instructed to overwrite this value before using it.
# Objective
When a global tracing subscriber has already been set, `LogPlugin` panics with an error message explaining this. However, if a global logger has already been set, it simply panics on an unwrap.
#6426 mentiones the panic and has been fixed by unique plugins, but the panic can still occur if a logger has been set through different means or multiple apps are created, as in #4934. The solution to that specific case isn't clear; this PR only fixes the missing error message.
## Solution
- ~add error message to panic~
- turn into warning
# Objective
- Reduce confusion around uniform bindings in materials. I've seen multiple people on discord get confused by it because it uses a struct that is named the same in the rust code and the wgsl code, but doesn't contain the same data. Also, the only reason this works is mostly by chance because the memory happens to align correctly.
## Solution
- Remove the confusing parts of the doc
## Notes
It's not super clear in the diff why this causes confusion, but essentially, the rust code defines a `CustomMaterial` struct with a color and a texture, but in the wgsl code the struct with the same name only contains the color. People are confused by it because the struct in wgsl doesn't need to be there.
You _can_ have complex structs on each side and the macro will even combine it for you if you reuse a binding index, but as it is now, this example seems to confuse more than help people.
# Objective
Fixes#6739
## Solution
Implement the required traits. They cannot be implemented for `Path` directly, since it is a dynamically-sized type.
# Objective
Fixes#6642
In a way that doesn't create any breaking changes, as a possible way to fix the above in a patch release.
## Solution
Don't actually remove font atlases when `max_font_atlases` is exceeded. Add a warning instead.
Keep `TextError::ExceedMaxTextAtlases` and `TextSettings` as-is so we don't break anything.
This is a bit of a cop-out, but the problems revealed by #6642 seem very challenging to fix properly.
Maybe follow up later with something more like https://github.com/rparrett/bevy/commits/remove-max-font-atlases later, if this is the direction we want to go.
## Note
See previous attempt at a "simple fix" that only solved some of the issues: #6666
# Objective
> Part of #6573
When serializing a `DynamicScene` we end up treating almost all non-value types as though their type data doesn't exist. This is because when creating the `DynamicScene` we call `Reflect::clone_value` on the components, which generates a Dynamic type for all non-value types.
What this means is that the `glam` types are treated as though their `ReflectSerialize` registrations don't exist. However, the deserializer _does_ pick up the registration and attempts to use that instead. This results in the deserializer trying to operate on "malformed" data, causing this error:
```
WARN bevy_asset::asset_server: encountered an error while loading an asset: Expected float
```
## Solution
Ideally, we should better handle the serialization of possibly-Dynamic types. However, this runs into issues where the `ReflectSerialize` expects the concrete type and not a Dynamic representation, resulting in a panic:
0aa4147af6/crates/bevy_reflect/src/type_registry.rs (L402-L413)
Since glam types are so heavily used in Bevy (specifically in `Transform` and `GlobalTransform`), it makes sense to just a quick fix in that enables them to be used properly in scenes while a proper solution is found.
This PR simply removes all `ReflectSerialize` and `ReflectDeserialize` registrations from the glam types that are reflected as structs.
---
## Changelog
- Remove `ReflectSerialize` and `ReflectDeserialize` registrations from most glam types
## Migration Guide
This PR removes `ReflectSerialize` and `ReflectDeserialize` registrations from most glam types. This means any code relying on either of those type data existing for those glam types will need to not do that.
This also means that some serialized glam types will need to be updated. For example, here is `Affine3A`:
```rust
// BEFORE
(
"glam::f32::affine3a::Affine3A": (1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0),
// AFTER
"glam::f32::affine3a::Affine3A": (
matrix3: (
x_axis: (
x: 1.0,
y: 0.0,
z: 0.0,
),
y_axis: (
x: 0.0,
y: 1.0,
z: 0.0,
),
z_axis: (
x: 0.0,
y: 0.0,
z: 1.0,
),
),
translation: (
x: 0.0,
y: 0.0,
z: 0.0,
),
)
)
```
# Objective
Many types in `bevy_render` implemented `Reflect` but were not registered.
## Solution
Register all types in `bevy_render` that impl `Reflect`.
This also registers additional dependent types (i.e. field types).
> Note: Adding these dependent types would not be needed using something like #5781😉
---
## Changelog
- Register missing `bevy_render` types in the `TypeRegistry`:
- `camera::RenderTarget`
- `globals::GlobalsUniform`
- `texture::Image`
- `view::ComputedVisibility`
- `view::Visibility`
- `view::VisibleEntities`
- Register additional dependent types:
- `view::ComputedVisibilityFlags`
- `Vec<Entity>`
# Objective
- Fixes https://github.com/bevyengine/bevy/issues/6603
## Solution
- `Task`s will cancel when dropped, but wait until they return Pending before they actually get canceled. That means that if a task panics, it's possible for that error to get propagated to the scope and the scope gets dropped, while scoped tasks in other threads are still running. This is a big problem since scoped task can hold life-timed values that are dropped as the scope is dropped leading to UB.
---
## Changelog
- changed `Scope` to use `FallibleTask` and await the cancellation of all remaining tasks when it's dropped.
# Objective
- Reverts #5730.
- Fixes#6173, fixes#6596.
## Solution
Remove the warning entirely.
## Changelog
You will no longer be spammed about
> Missed 31 `bevy_input:🐭:MouseMotion` events. Consider
reading from the `EventReader` more often (generally the best
solution) or calling Events::update() less frequently
(normally this is called once per frame). This problem is most
likely due to run criteria/fixed timesteps or consuming events
conditionally. See the Events documentation for
more information.
when you miss events. These warnings were often (but not always) a false positive. You can still check this manually by using `ManualEventReader::missed_events`
# Objective
Remove an obscure and inconsistent bit of API.
Simplify the `WorldChildBuilder` code.
No idea why this even exists.
An example of the removed API:
```rust
world.spawn_empty().with_children(|parent| {
parent.spawn_empty();
parent.push_children(&[some_entity]); // Does *not* add children to the parent.
// It's actually identical to:
parent.spawn_empty().push_children(&[some_entity]);
});
world.spawn_empty().with_children(|parent| {
// This just panics.
parent.push_children(&[some_entity]);
});
```
This exists only on `WorldChildBuilder`; `ChildBuilder` does not have this API.
Yeet.
## Migration Guide
Hierarchy editing methods such as `with_children` and `push_children` have been removed from `WorldChildBuilder`.
You can edit the hierarchy via `EntityMut` instead.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
Fixes#6713
Binary deserialization is failing for unit structs as well as structs with all ignored/skipped fields.
## Solution
Add a check for the number of possible fields in a struct before deserializing. If empty, don't attempt to deserialize any fields (as there will be none).
Note: ~~This does not apply to enums as they do not properly handle skipped fields (see #6721).~~ Enums still do not properly handle skipped fields, but I decided to include the logic for it anyways to account for `#[reflect(ignore)]`'d fields in the meantime.
---
## Changelog
- Fix bug where deserializing unit structs would fail for non-self-describing formats
# Objective
Consider the test
```rust
let cell = world.cell();
let _value_a = cell.resource_mut::<A>();
let _value_b = cell.resource_mut::<A>();
```
Currently, this will roughly execute
```rust
// first call
let value = unsafe {
self.world
.get_non_send_unchecked_mut_with_id(component_id)?
};
return Some(WorldBorrowMut::new(value, archetype_component_id, self.access)))
// second call
let value = unsafe {
self.world
.get_non_send_unchecked_mut_with_id(component_id)?
};
return Some(WorldBorrowMut::new(value, archetype_component_id, self.access)))
```
where `WorldBorrowMut::new` will panic if the resource is already borrowed.
This means, that `_value_a` will be created, the access checked (OK), then `value_b` will be created, and the access checked (`panic`).
For a moment, both `_value_a` and `_value_b` existed as `&mut T` to the same location, which is insta-UB as far as I understand it.
## Solution
Flip the order so that `WorldBorrowMut::new` first checks the access, _then_ fetches creates the value. To do that, we pass a `impl FnOnce() -> Mut<T>` instead of the `Mut<T>` directly:
```rust
let get_value = || unsafe {
self.world
.get_non_send_unchecked_mut_with_id(component_id)?
};
return Some(WorldBorrowMut::new(get_value, archetype_component_id, self.access)))
```
# Objective
- shaders defs can now have a `bool` or `int` value
- `#if SHADER_DEF <operator> 3`
- ok if `SHADER_DEF` is defined, has the correct type and pass the comparison
- `==`, `!=`, `>=`, `>`, `<`, `<=` supported
- `#SHADER_DEF` or `#{SHADER_DEF}`
- will be replaced by the value in the shader code
---
## Migration Guide
- replace `shader_defs.push(String::from("NAME"));` by `shader_defs.push("NAME".into());`
- if you used shader def `NO_STORAGE_BUFFERS_SUPPORT`, check how `AVAILABLE_STORAGE_BUFFER_BINDINGS` is now used in Bevy default shaders
# Objective
`add_node_edge` and `add_slot_edge` are fallible methods, but are always used with `.unwrap()`.
`input_node` is often unwrapped as well.
This points to having an infallible behaviour as default, with an alternative fallible variant if needed.
Improves readability and ergonomics.
## Solution
- Change `add_node_edge` and `add_slot_edge` to panic on error.
- Change `input_node` to panic on `None`.
- Add `try_add_node_edge` and `try_add_slot_edge` in case fallible methods are needed.
- Add `get_input_node` to still be able to get an `Option`.
---
## Changelog
### Added
- `try_add_node_edge`
- `try_add_slot_edge`
- `get_input_node`
### Changed
- `add_node_edge` is now infallible (panics on error)
- `add_slot_edge` is now infallible (panics on error)
- `input_node` now panics on `None`
## Migration Guide
Remove `.unwrap()` from `add_node_edge` and `add_slot_edge`.
For cases where the error was handled, use `try_add_node_edge` and `try_add_slot_edge` instead.
Remove `.unwrap()` from `input_node`.
For cases where the option was handled, use `get_input_node` instead.
Co-authored-by: Torstein Grindvik <52322338+torsteingrindvik@users.noreply.github.com>
# Objective
Fixes#4697. Hierarchical propagation of properties, currently only Transform -> GlobalTransform, can be a very expensive operation. Transform propagation is a strict dependency for anything positioned in world-space. In large worlds, this can take quite a bit of time, so limiting it to a single thread can result in poor CPU utilization as it bottlenecks the rest of the frame's systems.
## Solution
- Move transforms without a parent or a child (free-floating (Global)Transform) entities into a separate parallel system.
- Chunk the hierarchy based on the root entities and process it in parallel with `Query::par_for_each_mut`.
- Utilize the hierarchy's specific properties introduced in #4717 to allow for safe use of `Query::get_unchecked` on multiple threads. Assuming each child is unique in the hierarchy, it is impossible to have an aliased `&mut GlobalTransform` so long as we verify that the parent for a child is the same one propagated from.
---
## Changelog
Removed: `transform_propagate_system` is no longer `pub`.
Without this fix, piped systems containing exclusive systems fail to run, giving a runtime panic.
With this PR, running piped systems that contain exclusive systems now works.
## Explanation of the bug
This is because, unless overridden, the default implementation of `run` from the `System` trait simply calls `run_unsafe`. That is not valid for exclusive systems. They must always be called via `run`, as `run_unsafe` takes `&World` instead of `&mut World`.
Trivial reproduction example:
```rust
fn main() {
App::new()
.add_plugins(DefaultPlugins)
.add_system(exclusive.pipe(another))
.run();
}
fn exclusive(_world: &mut World) {}
fn another() {}
```
If you run this, you will get a panic 'Cannot run exclusive systems with a shared World reference' and the backtrace shows how bevy (correctly) tries to call the `run` method (because the system is exclusive), but it is the implementation from the `System` trait (because `PipeSystem` does not have its own), which calls `run_unsafe` (incorrect):
- 3: <bevy_ecs::system::system_piping::PipeSystem<SystemA,SystemB> as bevy_ecs::system::system::System>::run_unsafe
- 4: bevy_ecs::system::system::System::run
# Objective
Allow more use cases where the user may benefit from both `ExtractComponentPlugin` _and_ `UniformComponentPlugin`.
## Solution
Add an associated type to `ExtractComponent` in order to allow specifying the output component (or bundle).
Make `extract_component` return an `Option<_>` such that components can be extracted only when needed.
What problem does this solve?
`ExtractComponentPlugin` allows extracting components, but currently the output type is the same as the input.
This means that use cases such as having a settings struct which turns into a uniform is awkward.
For example we might have:
```rust
struct MyStruct {
enabled: bool,
val: f32
}
struct MyStructUniform {
val: f32
}
```
With the new approach, we can extract `MyStruct` only when it is enabled, and turn it into its related uniform.
This chains well with `UniformComponentPlugin`.
The user may then:
```rust
app.add_plugin(ExtractComponentPlugin::<MyStruct>::default());
app.add_plugin(UniformComponentPlugin::<MyStructUniform>::default());
```
This then saves the user a fair amount of boilerplate.
## Changelog
### Changed
- `ExtractComponent` can specify output type, and outputting is optional.
Co-authored-by: Torstein Grindvik <52322338+torsteingrindvik@users.noreply.github.com>
Add a method to rotate a transform to point towards a direction.
Also updated the docs to link to `forward` and `up` instead of mentioning local negative `Z` and local `Y`.
Unfortunately, links to methods don't work in rust-analyzer :(
Co-authored-by: Devil Ira <justthecooldude@gmail.com>
# Objective
Latest Release, "bevy 0.9" move the FrameCount updater into RenderPlugin, it leads to user who only run app with Core/Minimal Plugin cannot get the right number of FrameCount, it always return 0.
As for use cases like a server app, we don't want to add render dependencies to the app.
More detail in #6656
## Solution
- Move the `update_frame_count` into CorePlugin
# Objective
This add a ctor to `Box` to aid the creation of non-centred boxes. The PR adopts @rezural's work on PR #3322, taking into account the feedback on that PR from @james7132.
## Solution
`Box::from_corners()` creates a `Box` from two opposing corners and automatically determines the min and max extents to ensure that the `Box` is well-formed.
Co-authored-by: rezural <rezural@protonmail.com>
# Objective
- Bevy should be usable to create 'overlay' type apps, where the input is not captured by Bevy, but passed down/into a target app, or to allow passive displays/widgets etc.
## Solution
- the `winit:🪟:Window` already has a `set_cursor_hittest()` which basically does this for mouse input events, so I've exposed it (trying to copy the style laid out in the existing wrappings, and added a simple demo.
---
## Changelog
- Added `hittest` to `WindowAttributes`
- Added the `hittest`'s setters/getters
- Modified the `WindowBuilder`
- Modifed the `WindowDescriptor`'s `Default` impl.
- Added an example `cargo run --example fallthrough`
# Objective
Fixes#4884. `ComponentTicks` stores both added and changed ticks contiguously in the same 8 bytes. This is convenient when passing around both together, but causes half the bytes fetched from memory for the purposes of change detection to effectively go unused. This is inefficient when most queries (no filter, mutating *something*) only write out to the changed ticks.
## Solution
Split the storage for change detection ticks into two separate `Vec`s inside `Column`. Fetch only what is needed during iteration.
This also potentially also removes one blocker from autovectorization of dense queries.
EDIT: This is confirmed to enable autovectorization of dense queries in `for_each` and `par_for_each` where possible. Unfortunately `iter` has other blockers that prevent it.
### TODO
- [x] Microbenchmark
- [x] Check if this allows query iteration to autovectorize simple loops.
- [x] Clean up all of the spurious tuples now littered throughout the API
### Open Questions
- ~~Is `Mut::is_added` absolutely necessary? Can we not just use `Added` or `ChangeTrackers`?~~ It's optimized out if unused.
- ~~Does the fetch of the added ticks get optimized out if not used?~~ Yes it is.
---
## Changelog
Added: `Tick`, a wrapper around a single change detection tick.
Added: `Column::get_added_ticks`
Added: `Column::get_column_ticks`
Added: `SparseSet::get_added_ticks`
Added: `SparseSet::get_column_ticks`
Changed: `Column` now stores added and changed ticks separately internally.
Changed: Most APIs returning `&UnsafeCell<ComponentTicks>` now returns `TickCells` instead, which contains two separate `&UnsafeCell<Tick>` for either component ticks.
Changed: `Query::for_each(_mut)`, `Query::par_for_each(_mut)` will now leverage autovectorization to speed up query iteration where possible.
## Migration Guide
TODO
# Objective
Fix#6453.
## Solution
Use the solution mentioned in the issue by catching the unwind and dropping the error. Wrap the `executor.try_tick` calls with `std::catch::unwind`.
Ideally this would be moved outside of the hot loop, but the mut ref to the `spawned` future is not `UnwindSafe`.
This PR only addresses the bug, we can address the perf issues (should there be any) later.
# Objective
Fix#5149
## Solution
Instead of returning the **total count** of elements in the `QueryIter` in
`size_hint`, we return the **count of remaining elements**. This
Fixes#5149 even when #5148 gets merged.
- https://github.com/bevyengine/bevy/issues/5149
- https://github.com/bevyengine/bevy/pull/5148
---
## Changelog
- Fix partially consumed `QueryIter` and `QueryCombinationIter` having invalid `size_hint`
Co-authored-by: Nicola Papale <nicopap@users.noreply.github.com>
# Objective
Make core pipeline graphic nodes, including `BloomNode`, `FxaaNode`, `TonemappingNode` and `UpscalingNode` public.
This will allow users to construct their own render graphs with these build-in nodes.
## Solution
Make them public.
Also put node names into bevy's core namespace (`core_2d::graph::node`, `core_3d::graph::node`) which makes them consistent.
# Objective
Fix android touch events being flipped. Only removed test for android, don't have ios device to test with. Tested with emulator and physical device.
## Solution
Remove check, no longer needed with coordinate change in 0.9
# Objective
Delete `ImageMode`. It doesn't do anything except mislead people into thinking it controls the aspect ratio of images somehow.
Fixes#3933 and #6637
## Solution
Delete `ImageMode`
## Changelog
Removes the `ImageMode` enum.
Removes the `image_mode` field from `ImageBundle`
Removes the `With<ImageMode>` query filter from `image_node_system`
Renames `image_node_system` to` update_image_calculated_size_system`
# Objective
Fixes#6661
## Solution
Make `SceneSpawner::spawn_dynamic` return `InstanceId` like other functions there.
---
## Changelog
Make `SceneSpawner::spawn_dynamic` return `InstanceId` instead of `()`.
When I was upgrading to 0.9 noticed there were some changes to the timer, mainly the `TimerMode`. When switching from the old `is_repeating()` and `set_repeating()` to the new `mode()` and `set_mode()` noticed the docs still had the old description.
# Objective
BlobVec currently relies on a scratch piece of memory allocated at initialization to make a temporary copy of a component when using `swap_remove_and_{forget/drop}`. This is potentially suboptimal as it writes to a, well-known, but random part of memory instead of using the stack.
## Solution
As the `FIXME` in the file states, replace `swap_scratch` with a call to `swap_nonoverlapping::<u8>`. The swapped last entry is returned as a `OwnedPtr`.
In theory, this should be faster as the temporary swap is allocated on the stack, `swap_nonoverlapping` allows for easier vectorization for bigger types, and the same memory is used between the swap and the returned `OwnedPtr`.
# Objective
* Enable `Res` and `Query` parameter mutual exclusion
* Required for https://github.com/bevyengine/bevy/pull/5080
The `FilteredAccessSet::get_conflicts` methods didn't work properly with
`Res` and `ResMut` parameters. Because those added their access by using
the `combined_access_mut` method and directly modifying the global
access state of the FilteredAccessSet. This caused an inconsistency,
because get_conflicts assumes that ALL added access have a corresponding
`FilteredAccess` added to the `filtered_accesses` field.
In practice, that means that SystemParam that adds their access through
the `Access` returned by `combined_access_mut` and the ones that add
their access using the `add` method lived in two different universes. As
a result, they could never be mutually exclusive.
## Solution
This commit fixes it by removing the `combined_access_mut` method. This
ensures that the `combined_access` field of FilteredAccessSet is always
updated consistently with the addition of a filter. When checking for
filtered access, it is now possible to account for `Res` and `ResMut`
invalid access. This is currently not needed, but might be in the
future.
We add the `add_unfiltered_{read,write}` methods to replace previous
usages of `combined_access_mut`.
We also add improved Debug implementations on FixedBitSet so that their
meaning is much clearer in debug output.
---
## Changelog
* Fix `Res` and `Query` parameter never being mutually exclusive.
## Migration Guide
Note: this mostly changes ECS internals, but since the API is public, it is technically breaking:
* Removed `FilteredAccessSet::combined_access_mut`
* Replace _immutable_ usage of those by `combined_access`
* For _mutable_ usages, use the new `add_unfiltered_{read,write}` methods instead of `combined_access_mut` followed by `add_{read,write}`
# Objective
Make core types in ECS smaller. The column sparse set in Tables is never updated after creation.
## Solution
Create `ImmutableSparseSet` which removes the capacity fields in the backing vec's and the APIs for inserting or removing elements. Drops the size of the sparse set by 3 usizes (24 bytes on 64-bit systems)
## Followup
~~After #4809, Archetype's component SparseSet should be replaced with it.~~ This has been done.
---
## Changelog
Removed: `Table::component_capacity`
## Migration Guide
`Table::component_capacity()` has been removed as Tables do not support adding/removing columns after construction.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
Archetype is a deceptively large type in memory. It stores metadata about which components are in which storage in multiple locations, which is only used when creating new Archetypes while moving entities.
## Solution
Remove the redundant `Box<[ComponentId]>`s and iterate over the sparse set of component metadata instead. Reduces Archetype's size by 4 usizes (32 bytes on 64-bit systems), as well as the additional allocations for holding these slices.
It'd seem like there's a downside that the origin archetype has it's component metadata iterated over twice when creating a new archetype, but this change also removes the extra `Vec<ArchetypeComponentId>` allocations when creating a new archetype which may amortize out to a net gain here. This change likely negatively impacts creating new archetypes with a large number of components, but that's a cost mitigated by the fact that these archetypal relationships are cached in Edges and is incurred only once for each edge created.
## Additional Context
There are several other in-flight PRs that shrink Archetype:
- #4800 merges the entities and rows Vecs together (shaves off 24 bytes per archetype)
- #4809 removes unique_components and moves it to it's own dedicated storage (shaves off 72 bytes per archetype)
---
## Changelog
Changed: `Archetype::table_components` and `Archetype::sparse_set_components` return iterators instead of slices. `Archetype::new` requires iterators instead of parallel slices/vecs.
## Migration Guide
Do I still need to do this? I really hope people were not relying on the public facing APIs changed here.
# Objective
`ComputedVisibility` could afford to be smaller/faster. Optimizing the size and performance of operations on the component will positively benefit almost all extraction systems.
This was listed as one of the potential pieces of future work for #5310.
## Solution
Merge both internal booleans into a single `u8` bitflag field. Rely on bitmasks to evaluate local, hierarchical, and general visibility.
Pros:
- `ComputedVisibility::is_visible` should be a single bitmask test instead of two.
- `ComputedVisibility` is now only 1 byte. Should be able to fit 100% more per cache line when using dense iteration.
Cons:
- Harder to read.
- Setting individual values inside `ComputedVisiblity` require bitmask mutations.
This should be a non-breaking change. No public API was changed. The only publicly visible effect is that `ComputedVisibility` is now 1 byte instead of 2.
# Objective
Fixes#6594
## Solution
- `New` function for `Size` is now a `const` function :)
## Changelog
- `New` function for `Size` is now a `const` function
## Migration Guide
- Nothing has been changed
# Objective
In bevy 0.8 you could list all resources using `world.archetypes().resource().components()`. As far as I can tell the resource archetype has been replaced with the `Resources` storage, and it would be nice if it could be used to iterate over all resource component IDs as well.
## Solution
- add `fn Resources::iter(&self) -> impl Iterator<Item = (ComponentId, &ResourceData)>`
# Objective
- Derive Clone and Debug for `AssetPlugin`
- Make it possible to log asset server settings
- And get an owned instance if wrapping `AssetPlugin` in another plugin. See: 129224ef72/src/web_asset_plugin.rs (L45)
# Objective
> Part of #6573
`Children` was not being properly deserialized in scenes. This was due to a missing registration on `SmallVec<[Entity; 8]>`, which is used by `Children`.
## Solution
Register `SmallVec<[Entity; 8]>`.
---
## Changelog
- Registered `SmallVec<[Entity; 8]>`
# Objective
Fixes#6030, making ``serde`` optional.
## Solution
This was solved by making a ``serialize`` feature that can activate ``serde``, which is now optional.
When ``serialize`` is deactivated, the ``Plugin`` implementation for ``ScenePlugin`` does nothing.
Co-authored-by: Linus Käll <linus.kall.business@gmail.com>
# Objective
Currently, `Ptr` and `PtrMut` can only be constructed via unsafe code. This means that downgrading a reference to an untyped pointer is very cumbersome, despite being a very simple operation.
## Solution
Define conversions for easily and safely constructing untyped pointers. This is the non-owned counterpart to `OwningPtr::make`.
Before:
```rust
let ptr = unsafe { PtrMut::new(NonNull::from(&mut value).cast()) };
```
After:
```rust
let ptr = PtrMut::from(&mut value);
```
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
I needed a window which is always on top, to create a overlay app.
## Solution
expose the `always_on_top` property of winit in bevy's `WindowDescriptor` as a boolean flag
---
## Changelog
### Added
- add `WindowDescriptor.always_on_top` which configures a window to stay on top.
# Objective
`ScalingMode::Auto` for cameras only targets min_height and min_width, or as the docs say it `Use minimal possible viewport size while keeping the aspect ratio.`
But there is no ScalingMode that targets max_height and Max_width or `Use maximal possible viewport size while keeping the aspect ratio.`
## Solution
Added `ScalingMode::AutoMax` that does the exact opposite of `ScalingMode::Auto`
---
## Changelog
Renamed `ScalingMode::Auto` to `ScalingMode::AutoMin`.
## Migration Guide
just rename `ScalingMode::Auto` to `ScalingMode::AutoMin` if you are using it.
Co-authored-by: Lixou <82600264+DasLixou@users.noreply.github.com>
Add a method to get the focused window.
Use this instead of `WindowFocused` events in `close_on_esc`.
Seems that the OS/window manager might not always send focused events on application startup.
Sadly, not a fix for #5646.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
Fixes #3225, Allow for flippable UI Images
## Solution
Add flip_x and flip_y fields to UiImage, and swap the UV coordinates accordingly in ui_prepare_nodes.
## Changelog
* Changes UiImage to a struct with texture, flip_x, and flip_y fields.
* Adds flip_x and flip_y fields to ExtractedUiNode.
* Changes extract_uinodes to extract the flip_x and flip_y values from UiImage.
* Changes prepare_uinodes to swap the UV coordinates as required.
* Changes UiImage derefs to texture field accesses.
# Objective
Fixes#6615.
`BlobVec` does not respect alignment for zero-sized types, which results in UB whenever a ZST with alignment other than 1 is used in the world.
## Solution
Add the fn `bevy_ptr::dangling_with_align`.
---
## Changelog
+ Added the function `dangling_with_align` to `bevy_ptr`, which creates a well-aligned dangling pointer to a type whose alignment is not known at compile time.
# Objective
- Implements removal of entries from a `dyn Map`
- Fixes#6563
## Solution
- Adds a `remove` method to the `Map` trait which takes in a `&dyn Reflect` key and returns the value removed if it was present.
---
## Changelog
- Added `Map::remove`
## Migration Guide
- Implementors of `Map` will need to implement the `remove` method.
Co-authored-by: radiish <thesethskigamer@gmail.com>
# Objective
The [documentation for `ButtonSettingsError`](https://docs.rs/bevy/0.9.0/bevy/input/gamepad/enum.ButtonSettingsError.html) incorrectly describes the valid range of values as `0.0..=2.0`, probably because it was copied from `AxisSettingsError`. The actual range, as seen in the functions that return it and in its own `thiserror` description, is `0.0..=1.0`.
## Solution
Update the doc comments to reflect the correct range.
Co-authored-by: Sol Toder <ajaxgb@gmail.com>
# Objective
Fix#6548. Most of these methods were already made `const` in #5688. `Entity::to_bits` is the only one that remained.
## Solution
Make it const.
# Objective
- Fix#3606
- Fix#4579
- Fix#3380
## Solution
When running on a Linux machine with some AMD or Intel device, when calling
`surface.get_current_texture()`, ignore `wgpu::SurfaceError::Timeout` errors.
## Alternative
An alternative solution found in the `wgpu` examples is:
```rust
let frame = surface
.get_current_texture()
.or_else(|_| {
render_device.configure_surface(surface, &swap_chain_descriptor);
surface.get_current_texture()
})
.expect("Error reconfiguring surface");
window.swap_chain_texture = Some(TextureView::from(frame));
```
See: <94ce76391b/wgpu/examples/framework.rs (L362-L370)>
Veloren [handles the Timeout error the way this PR proposes to handle it](https://github.com/gfx-rs/wgpu/issues/1218#issuecomment-1092056971).
The reason I went with this PR's solution is that `configure_surface` seems to be quite an expensive operation, and it would run every frame with the wgpu framework solution, despite the fact it works perfectly fine without `configure_surface`.
I know this looks super hacky with the linux-specific line and the AMD check, but my understanding is that the `Timeout` occurrence is specific to a quirk of some AMD drivers on linux, and if otherwise met should be considered a bug.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
- Closes#5262
- Fix color banding caused by quantization.
## Solution
- Adds dithering to the tonemapping node from #3425.
- This is inspired by Godot's default "debanding" shader: https://gist.github.com/belzecue/
- Unlike Godot:
- debanding happens after tonemapping. My understanding is that this is preferred, because we are running the debanding at the last moment before quantization (`[f32, f32, f32, f32]` -> `f32`). This ensures we aren't biasing the dithering strength by applying it in a different (linear) color space.
- This code instead uses and reference the origin source, Valve at GDC 2015
## Additional Notes
Real time rendering to standard dynamic range outputs is limited to 8 bits of depth per color channel. Internally we keep everything in full 32-bit precision (`vec4<f32>`) inside passes and 16-bit between passes until the image is ready to be displayed, at which point the GPU implicitly converts our `vec4<f32>` into a single 32bit value per pixel, with each channel (rgba) getting 8 of those 32 bits.
### The Problem
8 bits of color depth is simply not enough precision to make each step invisible - we only have 256 values per channel! Human vision can perceive steps in luma to about 14 bits of precision. When drawing a very slight gradient, the transition between steps become visible because with a gradient, neighboring pixels will all jump to the next "step" of precision at the same time.
### The Solution
One solution is to simply output in HDR - more bits of color data means the transition between bands will become smaller. However, not everyone has hardware that supports 10+ bit color depth. Additionally, 10 bit color doesn't even fully solve the issue, banding will result in coherent bands on shallow gradients, but the steps will be harder to perceive.
The solution in this PR adds noise to the signal before it is "quantized" or resampled from 32 to 8 bits. Done naively, it's easy to add unneeded noise to the image. To ensure dithering is correct and absolutely minimal, noise is adding *within* one step of the output color depth. When converting from the 32bit to 8bit signal, the value is rounded to the nearest 8 bit value (0 - 255). Banding occurs around the transition from one value to the next, let's say from 50-51. Dithering will never add more than +/-0.5 bits of noise, so the pixels near this transition might round to 50 instead of 51 but will never round more than one step. This means that the output image won't have excess variance:
- in a gradient from 49 to 51, there will be a step between each band at 49, 50, and 51.
- Done correctly, the modified image of this gradient will never have a adjacent pixels more than one step (0-255) from each other.
- I.e. when scanning across the gradient you should expect to see:
```
|-band-| |-band-| |-band-|
Baseline: 49 49 49 50 50 50 51 51 51
Dithered: 49 50 49 50 50 51 50 51 51
Dithered (wrong): 49 50 51 49 50 51 49 51 50
```
You can see from above how correct dithering "fuzzes" the transition between bands to reduce distinct steps in color, without adding excess noise.
### HDR
The previous section (and this PR) assumes the final output is to an 8-bit texture, however this is not always the case. When Bevy adds HDR support, the dithering code will need to take the per-channel depth into account instead of assuming it to be 0-255. Edit: I talked with Rob about this and it seems like the current solution is okay. We may need to revisit once we have actual HDR final image output.
---
## Changelog
### Added
- All pipelines now support deband dithering. This is enabled by default in 3D, and can be toggled in the `Tonemapping` component in camera bundles. Banding is a graphical artifact created when the rendered image is crunched from high precision (f32 per color channel) down to the final output (u8 per channel in SDR). This results in subtle gradients becoming blocky due to the reduced color precision. Deband dithering applies a small amount of noise to the signal before it is "crunched", which breaks up the hard edges of blocks (bands) of color. Note that this does not add excess noise to the image, as the amount of noise is less than a single step of a color channel - just enough to break up the transition between color blocks in a gradient.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
- Make the many foxes not unnecessarily bright. Broken since #5666.
- Fixes#6528
## Solution
- In #5666 normalisation of normals was moved from the fragment stage to the vertex stage. However, it was not added to the vertex stage for skinned normals. The many foxes are skinned and their skinned normals were not unit normals. which made them brighter. Normalising the skinned normals fixes this.
---
## Changelog
- Fixed: Non-unit length skinned normals are now normalized.
This reverts commit 8429b6d6ca as discussed in #6522.
I tested that the game_menu example works as it should.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
Attempting to build `bevy_tasks` produces the following error:
```
error[E0599]: no method named `is_finished` found for struct `async_executor::Task` in the current scope
--> /[...]]/bevy/crates/bevy_tasks/src/task.rs:51:16
|
51 | self.0.is_finished()
| ^^^^^^^^^^^ method not found in `async_executor::Task<T>`
```
It looks like this was introduced along with `Task::is_finished`, which delegates to `async_task::Task::is_finished`. However, the latter was only introduced in `async-task` 4.2.0; `bevy_tasks` does not explicitly depend on `async-task` but on `async-executor` ^1.3.0, which in turn depends on `async-task` ^4.0.0.
## Solution
Add an explicit dependency on `async-task` ^4.2.0.
# Objective
Copy `send_event` and friends from `World` to `WorldCell`.
Clean up `bevy_winit` using `WorldCell::send_event`.
## Changelog
Added `send_event`, `send_event_default`, and `send_event_batch` to `WorldCell`.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
- it would be useful to inspect these structs using reflection
## Solution
- derive and register reflect
- Note that `#[reflect(Component)]` requires `Default` (or `FromWorld`) until #6060, so I implemented `Default` for `Tonemapping` with `is_enabled: false`
# Objective
The `load_internal_asset` macro is helpful when creating rendering plugins, but it doesn't support load binary assets (like those compiled as spir-v).
## Solution
Add a `load_internal_binary_asset` macro that use `include_bytes!`.
# Objective
Fixes#5393
## Solution
- Add padding to `GlobalsUniform` / `Globals` to make it 16-byte aligned.
Still not super clear on whether this is a `naga` thing or an `encase` thing or what. But now that we're offering `globals` up to users and #5393 is not just breaking an example, maybe we should do this sort of workaround?
* Move the despawn debug log from `World::despawn` to `EntityMut::despawn`.
* Move the despawn non-existent warning log from `Commands::despawn` to `World::despawn`.
This should make logging consistent regardless of which of the three `despawn` methods is used.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
Using `Reflect` we can easily switch between a specific reflection trait object, such as a `dyn Struct`, to a `dyn Reflect` object via `Reflect::as_reflect` or `Reflect::as_reflect_mut`.
```rust
fn do_something(value: &dyn Reflect) {/* ... */}
let foo: Box<dyn Struct> = Box::new(Foo::default());
do_something(foo.as_reflect());
```
However, there is no way to convert a _boxed_ reflection trait object to a `Box<dyn Reflect>`.
## Solution
Add a `Reflect::into_reflect` method which allows converting a boxed reflection trait object back into a boxed `Reflect` trait object.
```rust
fn do_something(value: Box<dyn Reflect>) {/* ... */}
let foo: Box<dyn Struct> = Box::new(Foo::default());
do_something(foo.into_reflect());
```
---
## Changelog
- Added `Reflect::into_reflect`
# Objective
Some render plugins, like [bevy-hikari](https://github.com/cryscan/bevy-hikari) require to set `CameraRenderGraph`. In order to switch between render graphs I need to insert a new `CameraRenderGraph` component. It's not very ergonomic.
## Solution
Add `CameraRenderGraph::set` like in [Name](https://docs.rs/bevy/latest/bevy/core/struct.Name.html).
---
## Changelog
### Added
- `CameraRenderGraph::set`.
# Objective
There is no way to gen an owned value of `Reflect`.
## Solution
Add it! This was originally a part of #6421, but @MrGVSV asked me to create a separate for it to implement reflect diffing.
---
## Changelog
### Added
- `Reflect::reflect_owned` to get an owned version of `Reflect`.
# Objective
`NodeBundle` contains an `image` field, which can be misleading, because if you do supply an image there, nothing will be shown to screen. You need to use an `ImageBundle` instead.
## Solution
* `image` (`UiImage`) field is removed from `NodeBundle`,
* extraction stage queries now make an optional query for `UiImage`, if one is not found, use the image handle that is used as a default by `UiImage`: c019a60b39/crates/bevy_ui/src/ui_node.rs (L464)
* touching up docs for `NodeBundle` to help guide what `NodeBundle` should be used for
* renamed `entity.rs` to `node_bundle.rs` as that gives more of a hint regarding the module's purpose
* separating `camera_config` stuff from the pre-made UI node bundles so that `node_bundle.rs` makes more sense as a module name.
# Objective
- adding a new `.register` should not overwrite old type data
- separate crates should both be able to register the same type
I ran into this while debugging why `register::<Handle<T>>` removed the `ReflectHandle` type data from a prior `register_asset_reflect`.
## Solution
- make `register` do nothing if called again for the same type
- I also removed some unnecessary duplicate registrations
# Objective
When an error causes `debug_checked_unreachable` to be called, the panic message unhelpfully points to the function definition instead of the place that caused the error.
## Solution
Add the `#[track_caller]` attribute in debug mode.
# Objective
Fixes: https://github.com/bevyengine/bevy/issues/6466
Summary: The UI Scaling example dynamically scales the UI which will dynamically allocate fonts to the font atlas surpassing the protective limit, throwing a panic.
## Solution
- Set TextSettings.allow_dynamic_font_size = true for the UI Scaling example. This is the ideal solution since the dynamic changes to the UI are not continuous yet still discrete.
- Update the panic text to reflect ui scaling as a potential cause
# Objective
Bevy UI (and third party plugins) currently have no good way to position themselves after all post processing effects. They currently use the tonemapping node, but this is not adequate if there is anything after tonemapping (such as FXAA).
## Solution
Add a logical `END_MAIN_PASS_POST_PROCESSING` RenderGraph node that main pass post processing effects position themselves before, and things like UIs can position themselves after.
# Objective
This PR fixes#5789, by enabling movable (and scalable) directional light shadow volumes.
## Solution
This PR changes `ExtractedDirectionalLight` to hold a copy of the `DirectionalLight` entity's `GlobalTransform`, instead of just a `direction` vector. This allows the shadow map volume (as defined by the light's `shadow_projection` field) to be transformed honoring translation _and_ scale transforms, and not just rotation.
It also augments the texel size calculation (used to determine the `shadow_normal_bias`) so that it now takes into account the upper bound of the x/y/z scale of the `GlobalTransform`.
This change makes the directional light extraction code more consistent with point and spot lights (that already use `transform`), and allows easily moving and scaling the shadow volume along with a player entity based on camera distance/angle, immediately enabling more real world use cases until we have a more sophisticated adaptive implementation, such as the one described in #3629.
**Note:** While it was previously possible to update the projection achieving a similar effect, depending on the light direction and distance to the origin, the fact that the shadow map camera was always positioned at the origin with a hardcoded `Vec3::Y` up value meant you would get sub-optimal or inconsistent/incorrect results.
---
## Changelog
### Changed
- `DirectionalLight` shadow volumes now honor translation and scale transforms
## Migration Guide
- If your directional lights were positioned at the origin and not scaled (the default, most common scenario) no changes are needed on your part; it just works as before;
- If you previously had a system for dynamically updating directional light shadow projections, you might now be able to simplify your code by updating the directional light entity's transform instead;
- In the unlikely scenario that a scene with directional lights that previously rendered shadows correctly has missing shadows, make sure your directional lights are positioned at (0, 0, 0) and are not scaled to a size that's too large or too small.
Co-authored-by: JMS55 <47158642+JMS55@users.noreply.github.com>
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
Co-authored-by: DGriffin91 <github@dgdigital.net>
`EntityMut::remove_children` does not call `self.update_location()` which is unsound.
Verified by adding the following assertion, which fails when running the tests.
```rust
let before = self.location();
self.update_location();
assert_eq!(before, self.location());
```
I also removed incorrect messages like "parent entity is not modified" and the unhelpful "Inserting a bundle in the children entities may change the parent entity's location if they were of the same archetype" which might lead people to think that's the *only* thing that can change the entity's location.
# Changelog
Added `EntityMut::world_scope`.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
Allow passing `Vec`s of glam vector types as vertex attributes.
Alternative to #4548 and #2719
Also used some macros to cut down on all the repetition.
# Migration Guide
Implementations of `From<Vec<[u16; 4]>>` and `From<Vec<[u8; 4]>>` for `VertexAttributeValues` have been removed.
I you're passing either `Vec<[u16; 4]>` or `Vec<[u8; 4]>` into `Mesh::insert_attribute` it will now require wrapping it with right the `VertexAttributeValues` enum variant.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
Closes#5934
Currently it is not possible to de/serialize data to non-self-describing formats using reflection.
## Solution
Add support for non-self-describing de/serialization using reflection.
This allows us to use binary formatters, like [`postcard`](https://crates.io/crates/postcard):
```rust
#[derive(Reflect, FromReflect, Debug, PartialEq)]
struct Foo {
data: String
}
let mut registry = TypeRegistry::new();
registry.register::<Foo>();
let input = Foo {
data: "Hello world!".to_string()
};
// === Serialize! === //
let serializer = ReflectSerializer::new(&input, ®istry);
let bytes: Vec<u8> = postcard::to_allocvec(&serializer).unwrap();
println!("{:?}", bytes); // Output: [129, 217, 61, 98, ...]
// === Deserialize! === //
let deserializer = UntypedReflectDeserializer::new(®istry);
let dynamic_output = deserializer
.deserialize(&mut postcard::Deserializer::from_bytes(&bytes))
.unwrap();
let output = <Foo as FromReflect>::from_reflect(dynamic_output.as_ref()).unwrap();
assert_eq!(expected, output); // OK!
```
#### Crates Tested
- ~~[`rmp-serde`](https://crates.io/crates/rmp-serde)~~ Apparently, this _is_ self-describing
- ~~[`bincode` v2.0.0-rc.1](https://crates.io/crates/bincode/2.0.0-rc.1) (using [this PR](https://github.com/bincode-org/bincode/pull/586))~~ This actually works for the latest release (v1.3.3) of [`bincode`](https://crates.io/crates/bincode) as well. You just need to be sure to use fixed-int encoding.
- [`postcard`](https://crates.io/crates/postcard)
## Future Work
Ideally, we would refactor the `serde` module, but I don't think I'll do that in this PR so as to keep the diff relatively small (and to avoid any painful rebases). This should probably be done once this is merged, though.
Some areas we could improve with a refactor:
* Split deserialization logic across multiple files
* Consolidate helper functions/structs
* Make the logic more DRY
---
## Changelog
- Add support for non-self-describing de/serialization using reflection.
Co-authored-by: Gino Valente <49806985+MrGVSV@users.noreply.github.com>
# Objective
- Adds a bloom pass for HDR-enabled Camera3ds.
- Supersedes (and all credit due to!) https://github.com/bevyengine/bevy/pull/3430 and https://github.com/bevyengine/bevy/pull/2876
## Solution
- A threshold is applied to isolate emissive samples, and then a series of downscale and upscaling passes are applied and composited together.
- Bloom is applied to 2d or 3d Cameras with hdr: true and a BloomSettings component.
---
## Changelog
- Added a `core_pipeline::bloom::BloomSettings` component.
- Added `BloomNode` that runs between the main pass and tonemapping.
- Added a `BloomPlugin` that is loaded as part of CorePipelinePlugin.
- Added a bloom example project.
Co-authored-by: JMS55 <47158642+JMS55@users.noreply.github.com>
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
Co-authored-by: DGriffin91 <github@dgdigital.net>
# Objective
Alternative to #6424Fixes#6226
Fixes spawning empty bundles
## Solution
Add `BundleComponentStatus` trait and implement it for `AddBundle` and a new `SpawnBundleStatus` type (which always returns an Added status). `write_components` is now generic on `BundleComponentStatus` instead of taking `AddBundle` directly. This means BundleSpawner can now avoid needing AddBundle from the Empty archetype, which means BundleSpawner no longer needs a reference to the original archetype.
In theory this cuts down on the work done in `write_components` when spawning, but I'm seeing no change in the spawn benchmarks.
# Objective
The UI pass in HDR breaks currently because the color attachment format does not match the HDR ViewTarget.
## Solution
Specialize the UI pipeline on "hdr-ness" and select the appropriate format (like we do in the other built in pipelines).
# Objective
- Fixes#4019
- Fix lighting of double-sided materials when using a negative scale
- The FlightHelmet.gltf model's hose uses a double-sided material. Loading the model with a uniform scale of -1.0, and comparing against Blender, it was identified that negating the world-space tangent, bitangent, and interpolated normal produces incorrect lighting. Discussion with Morten Mikkelsen clarified that this is both incorrect and unnecessary.
## Solution
- Remove the code that negates the T, B, and N vectors (the interpolated world-space tangent, calculated world-space bitangent, and interpolated world-space normal) when seeing the back face of a double-sided material with negative scale.
- Negate the world normal for a double-sided back face only when not using normal mapping
### Before, on `main`, flipping T, B, and N
<img width="932" alt="Screenshot 2022-08-22 at 15 11 53" src="https://user-images.githubusercontent.com/302146/185965366-f776ff2c-cfa1-46d1-9c84-fdcb399c273c.png">
### After, on this PR
<img width="932" alt="Screenshot 2022-08-22 at 15 12 11" src="https://user-images.githubusercontent.com/302146/185965420-8be493e2-3b1a-4188-bd13-fd6b17a76fe7.png">
### Double-sided material without normal maps
https://user-images.githubusercontent.com/302146/185988113-44a384e7-0b55-4946-9b99-20f8c803ab7e.mp4
---
## Changelog
- Fixed: Lighting of normal-mapped, double-sided materials applied to models with negative scale
- Fixed: Lighting and shadowing of back faces with no normal-mapping and a double-sided material
## Migration Guide
`prepare_normal` from the `bevy_pbr::pbr_functions` shader import has been reworked.
Before:
```rust
pbr_input.world_normal = in.world_normal;
pbr_input.N = prepare_normal(
pbr_input.material.flags,
in.world_normal,
#ifdef VERTEX_TANGENTS
#ifdef STANDARDMATERIAL_NORMAL_MAP
in.world_tangent,
#endif
#endif
in.uv,
in.is_front,
);
```
After:
```rust
pbr_input.world_normal = prepare_world_normal(
in.world_normal,
(material.flags & STANDARD_MATERIAL_FLAGS_DOUBLE_SIDED_BIT) != 0u,
in.is_front,
);
pbr_input.N = apply_normal_mapping(
pbr_input.material.flags,
pbr_input.world_normal,
#ifdef VERTEX_TANGENTS
#ifdef STANDARDMATERIAL_NORMAL_MAP
in.world_tangent,
#endif
#endif
in.uv,
);
```
# Objective
Replace `WorldQueryGats` trait with actual gats
## Solution
Replace `WorldQueryGats` trait with actual gats
---
## Changelog
- Replaced `WorldQueryGats` trait with actual gats
## Migration Guide
- Replace usage of `WorldQueryGats` assoc types with the actual gats on `WorldQuery` trait
Respect mipmap_filter when create ImageDescriptor with linear()/nearest()
# Objective
Fixes#6348
## Migration Guide
This PR changes default `ImageSettings` and may lead to unexpected behaviour for existing projects with mipmapped textures. Users should provide custom `ImageSettings` resource with `mipmap_filter=FilterMode::Nearest` if they want to keep old behaviour.
Co-authored-by: Yakov Borevich <j.borevich@gmail.com>
# Objective
Currently we are limiting the amount of direction lights in a scene to one.
## Solution
Increase the amount of direction lights from 1 to 10.
This still is not a perfect solution, but should unblock many use cases.
We could probably just store the directional lights similar to the point lights in an storage buffer, allowing for an variable amount of directional lights.
Co-authored-by: Kurt Kühnert <51823519+Ku95@users.noreply.github.com>
# Objective
Right now, the `TaskPool` implementation allows panics to permanently kill worker threads upon panicking. This is currently non-recoverable without using a `std::panic::catch_unwind` in every scheduled task. This is poor ergonomics and even poorer developer experience. This is exacerbated by #2250 as these threads are global and cannot be replaced after initialization.
Removes the need for temporary fixes like #4998. Fixes#4996. Fixes#6081. Fixes#5285. Fixes#5054. Supersedes #2307.
## Solution
The current solution is to wrap `Executor::run` in `TaskPool` with a `catch_unwind`, and discarding the potential panic. This was taken straight from [smol](404c7bcc0a/src/spawn.rs (L44))'s current implementation. ~~However, this is not entirely ideal as:~~
- ~~the signaled to the awaiting task. We would need to change `Task<T>` to use `async_task::FallibleTask` internally, and even then it doesn't signal *why* it panicked, just that it did.~~ (See below).
- ~~no error is logged of any kind~~ (See below)
- ~~it's unclear if it drops other tasks in the executor~~ (it does not)
- ~~This allows the ECS parallel executor to keep chugging even though a system's task has been dropped. This inevitably leads to deadlock in the executor.~~ Assuming we don't catch the unwind in ParallelExecutor, this will naturally kill the main thread.
### Alternatives
A final solution likely will incorporate elements of any or all of the following.
#### ~~Log and Ignore~~
~~Log the panic, drop the task, keep chugging. This only addresses the discoverability of the panic. The process will continue to run, probably deadlocking the executor. tokio's detatched tasks operate in this fashion.~~
Panics already do this by default, even when caught by `catch_unwind`.
#### ~~`catch_unwind` in `ParallelExecutor`~~
~~Add another layer catching system-level panics into the `ParallelExecutor`. How the executor continues when a core dependency of many systems fails to run is up for debate.~~
`async_task::Task` bubbles up panics already, this will transitively push panics all the way to the main thread.
#### ~~Emulate/Copy `tokio::JoinHandle` with `Task<T>`~~
~~`tokio::JoinHandle<T>` bubbles up the panic from the underlying task when awaited. This can be transitively applied across other APIs that also use `Task<T>` like `Query::par_for_each` and `TaskPool::scope`, bubbling up the panic until it's either caught or it reaches the main thread.~~
`async_task::Task` bubbles up panics already, this will transitively push panics all the way to the main thread.
#### Abort on Panic
The nuclear option. Log the error, abort the entire process on any thread in the task pool panicking. Definitely avoids any additional infrastructure for passing the panic around, and might actually lead to more efficient code as any unwinding is optimized out. However gives the developer zero options for dealing with the issue, a seemingly poor choice for debuggability, and prevents graceful shutdown of the process. Potentially an option for handling very low-level task management (a la #4740). Roughly takes the shape of:
```rust
struct AbortOnPanic;
impl Drop for AbortOnPanic {
fn drop(&mut self) {
abort!();
}
}
let guard = AbortOnPanic;
// Run task
std::mem::forget(AbortOnPanic);
```
---
## Changelog
Changed: `bevy_tasks::TaskPool`'s threads will no longer terminate permanently when a task scheduled onto them panics.
Changed: `bevy_tasks::Task` and`bevy_tasks::Scope` will propagate panics in the spawned tasks/scopes to the parent thread.
# Objective
Add consistent UI rendering and interaction where deep nodes inside two different hierarchies will never render on top of one-another by default and offer an escape hatch (z-index) for nodes to change their depth.
## The problem with current implementation
The current implementation of UI rendering is broken in that regard, mainly because [it sets the Z value of the `Transform` component based on a "global Z" space](https://github.com/bevyengine/bevy/blob/main/crates/bevy_ui/src/update.rs#L43) shared by all nodes in the UI. This doesn't account for the fact that each node's final `GlobalTransform` value will be relative to its parent. This effectively makes the depth unpredictable when two deep trees are rendered on top of one-another.
At the moment, it's also up to each part of the UI code to sort all of the UI nodes. The solution that's offered here does the full sorting of UI node entities once and offers the result through a resource so that all systems can use it.
## Solution
### New ZIndex component
This adds a new optional `ZIndex` enum component for nodes which offers two mechanism:
- `ZIndex::Local(i32)`: Overrides the depth of the node relative to its siblings.
- `ZIndex::Global(i32)`: Overrides the depth of the node relative to the UI root. This basically allows any node in the tree to "escape" the parent and be ordered relative to the entire UI.
Note that in the current implementation, omitting `ZIndex` on a node has the same result as adding `ZIndex::Local(0)`. Additionally, the "global" stacking context is essentially a way to add your node to the root stacking context, so using `ZIndex::Local(n)` on a root node (one without parent) will share that space with all nodes using `Index::Global(n)`.
### New UiStack resource
This adds a new `UiStack` resource which is calculated from both hierarchy and `ZIndex` during UI update and contains a vector of all node entities in the UI, ordered by depth (from farthest from camera to closest). This is exposed publicly by the bevy_ui crate with the hope that it can be used for consistent ordering and to reduce the amount of sorting that needs to be done by UI systems (i.e. instead of sorting everything by `global_transform.z` in every system, this array can be iterated over).
### New z_index example
This also adds a new z_index example that showcases the new `ZIndex` component. It's also a good general demo of the new UI stack system, because making this kind of UI was very broken with the old system (e.g. nodes would render on top of each other, not respecting hierarchy or insert order at all).
---
## Changelog
- Added the `ZIndex` component to bevy_ui.
- Added the `UiStack` resource to bevy_ui, and added implementation in a new `stack.rs` module.
- Removed the previous Z updating system from bevy_ui, because it was replaced with the above.
- Changed bevy_ui rendering to use UiStack instead of z ordering.
- Changed bevy_ui focus/interaction system to use UiStack instead of z ordering.
- Added a new z_index example.
## ZIndex demo
Here's a demo I wrote to test these features
https://user-images.githubusercontent.com/1060971/188329295-d7beebd6-9aee-43ab-821e-d437df5dbe8a.mp4
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
This reverts commit 53d387f340.
# Objective
Reverts #6448. This didn't have the intended effect: we're now getting bevy::prelude shown in the docs again.
Co-authored-by: Alejandro Pascual <alejandro.pascual.pozo@gmail.com>
# Objective
- Right now re-exports are completely hidden in prelude docs.
- Fixes#6433
## Solution
- We could show the re-exports without inlining their documentation.
# Objective
Fixes#6059, changing all incorrect occurrences of ``id`` in the ``entity`` module to ``index``:
* struct level documentation,
* ``id`` struct field,
* ``id`` method and its documentation.
## Solution
Renaming and verifying using CI.
Co-authored-by: Edvin Kjell <43633999+Edwox@users.noreply.github.com>
# Objective
In some scenarios it can be useful to check if a task has been finished without polling it. I added a function called `is_finished` to check if a task has been finished.
## Solution
Since `async_task` supports it out of the box, it is just a simple wrapper function.
---
# Objective
- Add post processing passes for FXAA (Fast Approximate Anti-Aliasing)
- Add example comparing MSAA and FXAA
## Solution
When the FXAA plugin is added, passes for FXAA are inserted between the main pass and the tonemapping pass. Supports using either HDR or LDR output from the main pass.
---
## Changelog
- Add a new FXAANode that runs after the main pass when the FXAA plugin is added.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
For `derive(WorldQuery)`, there are three structs generated, `Item`, `Fetch` and `State`.
These inherit the visibility of the derived structure, thus `#![warn(missing_docs)]` would
warn about missing documentation for these structures.
- [ ] I'd like some advice on what to write here, as I personally don't really understand `Fetch` nor `State`.
# Objective
Currently, `bevy_dynamic_plugin` simply panics on error. This makes it impossible to handle failures in applications that use this feature.
For example, I'd like to build an optional expansion for my game, that may not be distributed to all users. I want to use `bevy_dynamic_plugin` for loading it. I want my game to try to load it on startup, but continue without it if it cannot be loaded.
## Solution
- Make the `dynamically_load_plugin` function return a `Result`, so it can gracefully return loading errors.
- Create an error enum type, to provide useful information about the kind of error. This adds `thiserror` to the dependencies of `bevy_dynamic_plugin`, but that dependency is already used in other parts of bevy (such as `bevy_asset`), so not a big deal.
I chose not to change the behavior of the builder method in the App extension trait. I kept it as panicking. There is no clean way (that I'm aware of) to make a builder-style API that has fallible methods. So it is either a panic or a warning. I feel the panic is more appropriate.
---
## Changelog
### Changed
- `bevy_dynamic_plugin::dynamically_load_plugin` now returns `Result` instead of panicking, to allow for error handling
# Objective
* Add benchmarks for `Query::get_many`.
* Speed up `Query::get_many`.
## Solution
Previously, `get_many` and `get_many_mut` used the method `array::map`, which tends to optimize very poorly. This PR replaces uses of that method with loops.
## Benchmarks
| Benchmark name | Execution time | Change from this PR |
|--------------------------------------|----------------|---------------------|
| query_get_many_2/50000_calls_table | 1.3732 ms | -24.967% |
| query_get_many_2/50000_calls_sparse | 1.3826 ms | -24.572% |
| query_get_many_5/50000_calls_table | 2.6833 ms | -30.681% |
| query_get_many_5/50000_calls_sparse | 2.9936 ms | -30.672% |
| query_get_many_10/50000_calls_table | 5.7771 ms | -36.950% |
| query_get_many_10/50000_calls_sparse | 7.4345 ms | -36.987% |
# Objective
Add documentation `#[world_query(ignore)]`. Fixes#6283.
---
I've only described it's behavior so far (which appears to be the same as with `system_param`). Is there another use-case for this besides with `PhantomData`? I could only find a single usage of this construct on GitHub, which is [here](ffcb816927/bevy/examples/ecs/custom_query_param.rs (L102)).
I was also wondering if it would make sense to add a usage example to the `custom_query_example`? 🤔 That's why it's currently still in there.
Co-authored-by: Lucas Jenß <243719+x3ro@users.noreply.github.com>
# Objective
`bevy_core` is missing a feature corresponding to the `serialize` feature on the `bevy` crate. Similar to #6378 and https://github.com/bevyengine/bevy/pull/6379 to serialize `Name` easily.
## Solution
Add this feature and hand-written serialization for `Name` (to avoid storing `hash` field).
---
## Changelog
### Added
* `Serialize` and `Deserialize` derives for `Name` under `serialize` feature.
# Objective
Post processing effects cannot read and write to the same texture. Currently they must own their own intermediate texture and redundantly copy from that back to the main texture. This is very inefficient.
Additionally, working with ViewTarget is more complicated than it needs to be, especially when working with HDR textures.
## Solution
`ViewTarget` now stores two copies of the "main texture". It uses an atomic value to track which is currently the "main texture" (this interior mutability is necessary to accommodate read-only RenderGraph execution).
`ViewTarget` now has a `post_process_write` method, which will return a source and destination texture. Each call to this method will flip between the two copies of the "main texture".
```rust
let post_process = render_target.post_process_write();
let source_texture = post_process.source;
let destination_texture = post_process.destination;
```
The caller _must_ read from the source texture and write to the destination texture, as it is assumed that the destination texture will become the new "main texture".
For simplicity / understandability `ViewTarget` is now a flat type. "hdr-ness" is a property of the `TextureFormat`. The internals are fully private in the interest of providing simple / consistent apis. Developers can now easily access the main texture by calling `view_target.main_texture()`.
HDR ViewTargets no longer have an "ldr texture" with `TextureFormat::bevy_default`. They _only_ have their two "hdr" textures. This simplifies the mental model. All we have is the "currently active hdr texture" and the "other hdr texture", which we flip between for post processing effects.
The tonemapping node has been rephrased to use this "post processing pattern". The blit pass has been removed, and it now only runs a pass when HDR is enabled. Notably, both the input and output texture are assumed to be HDR. This means that tonemapping behaves just like any other "post processing effect". It could theoretically be moved anywhere in the "effect chain" and continue to work.
In general, I think these changes will make the lives of people making post processing effects much easier. And they better position us to start building higher level / more structured "post processing effect stacks".
---
## Changelog
- `ViewTarget` now stores two copies of the "main texture". Calling `ViewTarget::post_process_write` will flip between copies of the main texture.
# Objective
- `ReflectDefault` can be used to create default values for reflected types
- `std` primitives that are `Default`-constructable should register `ReflectDefault`
## Solution
- register `ReflectDefault`
# Objective
- Fixes #6311
- Make it clearer what should be done in the example (close the Bevy app window)
## Solution
- Remove the second windowed Bevy App [since winit does not support this](https://github.com/rust-windowing/winit/blob/v0.27.4/src/event_loop.rs#L82-L83)
- Add title to the Bevy window asking the user to close it
This is more of a quick fix to have a working example. It would be nicer if we had a small real usecase for this functionality.
Another alternativ that I tried out: If we want to showcase a second Bevy app as it was before, we could still do this as long as one of them does not have a window. But I don't see how this is helpful in the context of the example, so I stuck with only one Bevy app and a simple print afterwards.
# Objective
Entities are unique, however, this is not reflected in the scene format. Currently, entities are stored in a list where a user could inadvertently create a duplicate of the same entity.
## Solution
Switch from the list representation to a map representation for entities.
---
## Changelog
* The `entities` field in the scene format is now a map of entity ID to entity data
## Migration Guide
The scene format now stores its collection of entities in a map rather than a list:
```rust
// OLD
(
entities: [
(
entity: 12,
components: {
"bevy_transform::components::transform::Transform": (
translation: (
x: 0.0,
y: 0.0,
z: 0.0
),
rotation: (0.0, 0.0, 0.0, 1.0),
scale: (
x: 1.0,
y: 1.0,
z: 1.0
),
),
},
),
],
)
// NEW
(
entities: {
12: (
components: {
"bevy_transform::components::transform::Transform": (
translation: (
x: 0.0,
y: 0.0,
z: 0.0
),
rotation: (0.0, 0.0, 0.0, 1.0),
scale: (
x: 1.0,
y: 1.0,
z: 1.0
),
),
},
),
},
)
```
# Objective
- Make it impossible to add a plugin twice
- This is going to be more a risk for plugins with configurations, to avoid things like `App::new().add_plugins(DefaultPlugins).add_plugin(ImagePlugin::default_nearest())`
## Solution
- Panic when a plugin is added twice
- It's still possible to mark a plugin as not unique by overriding `is_unique`
- ~~Simpler version of~~ #3988 (not simpler anymore because of how `PluginGroupBuilder` implements `PluginGroup`)
# Objective
- Bevy main crashs on Safari mobile
- On Safari mobile, calling winit_window.set_cursor_grab(true) fails as the API is not implemented (as there is no cursor on Safari mobile, the api doesn't make sense there). I don't know about other mobile browsers
## Solution
- Do not call the api to release cursor grab on window creation, as the cursor is not grabbed anyway at this point
- This is #3617 which was lost in #6218
# Objective
Fixes#6378
`bevy_transform` is missing a feature corresponding to the `serialize` feature on the `bevy` crate.
## Solution
Adds a `serialize` feature to `bevy_transform`.
Derives `serde::Serialize` and `Deserialize` when feature is enabled.
# Objective
Currently toggling an `AudioSink` (for example from a game menu) requires writing
```rs
if sink.is_paused() {
sink.play();
} else {
sink.pause();
}
```
It would be nicer if we could reduce this down to a single line
```rs
sink.toggle();
```
## Solution
Add an `AudioSink::toggle` method which does exactly that.
---
## Changelog
- Added `AudioSink::toggle` which can be used to toggle state of a sink.
# Objective
Add methods to `Query<&Children>` and `Query<&Parent>` to iterate over descendants and ancestors, respectively.
## Changelog
* Added extension trait for `Query` in `bevy_hierarchy`, `HierarchyQueryExt`
* Added method `iter_descendants` to `Query<&Children>` via `HierarchyQueryExt` for iterating over the descendants of an entity.
* Added method `iter_ancestors` to `Query<&Parent>` via `HierarchyQueryExt` for iterating over the ancestors of an entity.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
- Freeing unused memory held by visible entities
- Fixed comment style
# Objective
With Rust 1.56 it's possible to shrink vectors to a specified capacity. Visibility system had a comment before asking for that feature to free unused memory by a vector if its capacity is two times larger than the length.
## Solution
Shrinking the vector of visible entities to the nearest power of 2 elements next to `len()`, if capacity exceeds it more than two times.
# Objective
- Time have `Reflect`, but doesn't have `FromReflect`.
## Solution
- Add it for `Timer`, `Stopwatch` and `TimerMode`.
---
## Changelog
### Added
* `FromReflect` derive for `Timer`, `Stopwatch` and `TimerMode`.
# Objective
Bevy still has many instances of using single-tuples `(T,)` to create a bundle. Due to #2975, this is no longer necessary.
## Solution
Search for regex `\(.+\s*,\)`. This should have found every instance.
# Objective
Fix the soundness issue outlined in #5866. In short the problem is that `query.to_readonly().get_component_mut::<T>()` can provide unsound mutable access to the component. This PR is an alternative to just removing the offending api. Given that `to_readonly` is a useful tool, I think this approach is a preferable short term solution. Long term I think theres a better solution out there, but we can find that on its own time.
## Solution
Add what amounts to a "dirty flag" that marks Queries that have been converted to their read-only variant via `to_readonly` as dirty. When this flag is set to true, `get_component_mut` will fail with an error, preventing the unsound access.
# Objective
Following discussion on #3536 and #3522, `Handle::as_weak()` takes a type `U`, reinterpreting the handle as of another asset type while keeping the same ID. This is mainly used today in font atlas code. This PR does two things:
- Rename the method to `cast_weak()` to make its intent more clear
- Actually change the type uuid in the handle if it's not an asset path variant.
## Migration Guide
- Rename `Handle::as_weak` uses to `Handle::cast_weak`
The method now properly sets the associated type uuid if the handle is a direct reference (e.g. not a reference to an `AssetPath`), so adjust you code accordingly if you relied on the previous behavior.
# Objective
Currently for entities we serialize only `id`. But this is not very expected behavior. For example, in networking, when the server sends its state, it contains entities and components. On the client, I create new objects and map them (using `EntityMap`) to those received from the server (to know which one matches which). And if `generation` field is missing, this mapping can be broken. Example:
1. Server sends an entity `Entity{ id: 2, generation: 1}` with components.
2. Client puts the received entity in a map and create a new entity that maps to this received entity. The new entity have different `id` and `generation`. Let's call it `Entity{ id: 12, generation: 4}`.
3. Client sends a command for `Entity{ id: 12, generation: 4}`. To do so, it maps local entity to the one from server. But `generation` field is 0 because it was omitted for serialization on the server. So it maps to `Entity{ id: 2, generation: 0}`.
4. Server receives `Entity{ id: 2, generation: 0}` which is invalid.
In my game I worked around it by [writing custom serialization](https://github.com/dollisgame/dollis/blob/master/src/core/network/entity_serde.rs) and using `serde(with = "...")`. But it feels like a bad default to me.
Using `Entity` over a custom `NetworkId` also have the following advantages:
1. Re-use `MapEntities` trait to map `Entity`s in replicated components.
2. Instead of server `Entity <-> NetworkId ` and `Entity <-> NetworkId`, we map entities only on client.
3. No need to handling uniqueness. It's a rare case, but makes things simpler. For example, I don't need to query for a resource to create an unique ID.
Closes#6143.
## Solution
Use default serde impls. If anyone want to avoid wasting memory on `generation`, they can create a new type that holds `u32`. This is what Bevy do for [DynamicEntity](https://docs.rs/bevy/latest/bevy/scene/struct.DynamicEntity.html) to serialize scenes. And I don't see any use case to serialize an entity id expect this one.
---
## Changelog
### Changed
- Entity now serializes / deserializes `generation` field.
## Migration Guide
- Entity now fully serialized. If you want to serialze only `id`, as it was before, you can create a new type that wraps `u32`.
# Objective
- fix new clippy lints before they get stable and break CI
## Solution
- run `clippy --fix` to auto-fix machine-applicable lints
- silence `clippy::should_implement_trait` for `fn HandleId::default<T: Asset>`
## Changes
- always prefer `format!("{inline}")` over `format!("{}", not_inline)`
- prefer `Box::default` (or `Box::<T>::default` if necessary) over `Box::new(T::default())`
# Objective
^ enable this
Concretely, I need to
- list all handle ids for an asset type
- fetch the asset as `dyn Reflect`, given a `HandleUntyped`
- when encountering a `Handle<T>`, find out what asset type that handle refers to (`T`'s type id) and turn the handle into a `HandleUntyped`
## Solution
- add `ReflectAsset` type containing function pointers for working with assets
```rust
pub struct ReflectAsset {
type_uuid: Uuid,
assets_resource_type_id: TypeId, // TypeId of the `Assets<T>` resource
get: fn(&World, HandleUntyped) -> Option<&dyn Reflect>,
get_mut: fn(&mut World, HandleUntyped) -> Option<&mut dyn Reflect>,
get_unchecked_mut: unsafe fn(&World, HandleUntyped) -> Option<&mut dyn Reflect>,
add: fn(&mut World, &dyn Reflect) -> HandleUntyped,
set: fn(&mut World, HandleUntyped, &dyn Reflect) -> HandleUntyped,
len: fn(&World) -> usize,
ids: for<'w> fn(&'w World) -> Box<dyn Iterator<Item = HandleId> + 'w>,
remove: fn(&mut World, HandleUntyped) -> Option<Box<dyn Reflect>>,
}
```
- add `ReflectHandle` type relating the handle back to the asset type and providing a way to create a `HandleUntyped`
```rust
pub struct ReflectHandle {
type_uuid: Uuid,
asset_type_id: TypeId,
downcast_handle_untyped: fn(&dyn Any) -> Option<HandleUntyped>,
}
```
- add the corresponding `FromType` impls
- add a function `app.register_asset_reflect` which is supposed to be called after `.add_asset` and registers `ReflectAsset` and `ReflectHandle` in the type registry
---
## Changelog
- add `ReflectAsset` and `ReflectHandle` types, which allow code to use reflection to manipulate arbitrary assets without knowing their types at compile time
fixes https://github.com/bevyengine/bevy/issues/5944
Uses the second solution:
> 2. keep track of the old viewport in the computed_state, and if camera.viewport != camera.computed_state.old_viewport, then update the projection. This is more reliable, but needs to store two UVec2s more in the camera (probably not a big deal).
# Objective
Clean up code surrounding fetch by pulling out the common parts into the iteration code.
## Solution
Merge `Fetch::table_fetch` and `Fetch::archetype_fetch` into a single API: `Fetch::fetch(&mut self, entity: &Entity, table_row: &usize)`. This provides everything any fetch requires to internally decide which storage to read from and get the underlying data. All of these functions are marked as `#[inline(always)]` and the arguments are passed as references to attempt to optimize out the argument that isn't being used.
External to `Fetch`, Query iteration has been changed to keep track of the table row and entity outside of fetch, which moves a lot of the expensive bookkeeping `Fetch` structs had previously done internally into the outer loop.
~~TODO: Benchmark, docs~~ Done.
---
## Changelog
Changed: `Fetch::table_fetch` and `Fetch::archetype_fetch` have been merged into a single `Fetch::fetch` function.
## Migration Guide
TODO
Co-authored-by: Brian Merchant <bhmerchang@gmail.com>
Co-authored-by: Saverio Miroddi <saverio.pub2@gmail.com>
# Objective
Currently, `DynamicSceneBuilder` keeps track of entities via a `HashMap`. This has an unintended side-effect in that, when building the full `DynamicScene`, we aren't guaranteed any particular order.
In other words, inserting Entity A then Entity B can result in either `[A, B]` or `[B, A]`. This can be rather annoying when running tests on scenes generated via the builder as it will work sometimes but not other times. There's also the potential that this might unnecessarily clutter up VCS diffs for scene files (assuming they had an intentional order).
## Solution
Store `DynamicSceneBuilder`'s entities in a `Vec` rather than a `HashMap`.
---
## Changelog
* Stablized entity order in `DynamicSceneBuilder` (0.9.0-dev)
# Objective
Bevy's internal plugins have lots of execution-order ambiguities, which makes the ambiguity detection tool very noisy for our users.
## Solution
Silence every last ambiguity that can currently be resolved.
Each time an ambiguity is silenced, it is accompanied by a comment describing why it is correct. This description should be based on the public API of the respective systems. Thus, I have added documentation to some systems describing how they use some resources.
# Future work
Some ambiguities remain, due to issues out of scope for this PR.
* The ambiguity checker does not respect `Without<>` filters, leading to false positives.
* Ambiguities between `bevy_ui` and `bevy_animation` cannot be resolved, since neither crate knows that the other exists. We will need a general solution to this problem.
# Objective
- Fixes#5876 .
## Solution
- added pub use statements to re-export the following traits in bevy_audio: rodio::source::Source, rodio::Sample, rodio::cpal::Sample.
- rodio::cpal::Sample was re-exported as CpalSample to avoid naming conflict with rodio::Sample.
# Objective
Currently scenes define components using a list:
```rust
[
(
entity: 0,
components: [
{
"bevy_transform::components::transform::Transform": (
translation: (
x: 0.0,
y: 0.0,
z: 0.0
),
rotation: (0.0, 0.0, 0.0, 1.0),
scale: (
x: 1.0,
y: 1.0,
z: 1.0
),
),
},
{
"my_crate::Foo": (
text: "Hello World",
),
},
{
"my_crate::Bar": (
baz: 123,
),
},
],
),
]
```
However, this representation has some drawbacks (as pointed out by @Metadorius in [this](https://github.com/bevyengine/bevy/pull/4561#issuecomment-1202215565) comment):
1. Increased nesting and more characters (minor effect on overall size)
2. More importantly, by definition, entities cannot have more than one instance of any given component. Therefore, such data is best stored as a map— where all values are meant to have unique keys.
## Solution
Change `components` to store a map of components rather than a list:
```rust
[
(
entity: 0,
components: {
"bevy_transform::components::transform::Transform": (
translation: (
x: 0.0,
y: 0.0,
z: 0.0
),
rotation: (0.0, 0.0, 0.0, 1.0),
scale: (
x: 1.0,
y: 1.0,
z: 1.0
),
),
"my_crate::Foo": (
text: "Hello World",
),
"my_crate::Bar": (
baz: 123
),
},
),
]
```
#### Code Representation
This change only affects the scene format itself. `DynamicEntity` still stores its components as a list. The reason for this is that storing such data as a map is not really needed since:
1. The "key" of each value is easily found by just calling `Reflect::type_name` on it
2. We should be generating such structs using the `World` itself which upholds the one-component-per-entity rule
One could in theory create manually create a `DynamicEntity` with duplicate components, but this isn't something I think we should focus on in this PR. `DynamicEntity` can be broken in other ways (i.e. storing a non-component in the components list), and resolving its issues can be done in a separate PR.
---
## Changelog
* The scene format now uses a map to represent the collection of components rather than a list
## Migration Guide
The scene format now uses a map to represent the collection of components. Scene files will need to update from the old list format.
<details>
<summary>Example Code</summary>
```rust
// OLD
[
(
entity: 0,
components: [
{
"bevy_transform::components::transform::Transform": (
translation: (
x: 0.0,
y: 0.0,
z: 0.0
),
rotation: (0.0, 0.0, 0.0, 1.0),
scale: (
x: 1.0,
y: 1.0,
z: 1.0
),
),
},
{
"my_crate::Foo": (
text: "Hello World",
),
},
{
"my_crate::Bar": (
baz: 123,
),
},
],
),
]
// NEW
[
(
entity: 0,
components: {
"bevy_transform::components::transform::Transform": (
translation: (
x: 0.0,
y: 0.0,
z: 0.0
),
rotation: (0.0, 0.0, 0.0, 1.0),
scale: (
x: 1.0,
y: 1.0,
z: 1.0
),
),
"my_crate::Foo": (
text: "Hello World",
),
"my_crate::Bar": (
baz: 123
),
},
),
]
```
</details>
# Objective
Currently, Bevy only supports rendering to the current "surface texture format". This means that "render to texture" scenarios must use the exact format the primary window's surface uses, or Bevy will crash. This is even harder than it used to be now that we detect preferred surface formats at runtime instead of using hard coded BevyDefault values.
## Solution
1. Look up and store each window surface's texture format alongside other extracted window information
2. Specialize the upscaling pass on the current `RenderTarget`'s texture format, now that we can cheaply correlate render targets to their current texture format
3. Remove the old `SurfaceTextureFormat` and `AvailableTextureFormats`: these are now redundant with the information stored on each extracted window, and probably should not have been globals in the first place (as in theory each surface could have a different format).
This means you can now use any texture format you want when rendering to a texture! For example, changing the `render_to_texture` example to use `R16Float` now doesn't crash / properly only stores the red component:
Attempt to make features like bloom https://github.com/bevyengine/bevy/pull/2876 easier to implement.
**This PR:**
- Moves the tonemapping from `pbr.wgsl` into a separate pass
- also add a separate upscaling pass after the tonemapping which writes to the swap chain (enables resolution-independant rendering and post-processing after tonemapping)
- adds a `hdr` bool to the camera which controls whether the pbr and sprite shaders render into a `Rgba16Float` texture
**Open questions:**
- ~should the 2d graph work the same as the 3d one?~ it is the same now
- ~The current solution is a bit inflexible because while you can add a post processing pass that writes to e.g. the `hdr_texture`, you can't write to a separate `user_postprocess_texture` while reading the `hdr_texture` and tell the tone mapping pass to read from the `user_postprocess_texture` instead. If the tonemapping and upscaling render graph nodes were to take in a `TextureView` instead of the view entity this would almost work, but the bind groups for their respective input textures are already created in the `Queue` render stage in the hardcoded order.~ solved by creating bind groups in render node
**New render graph:**
<details>
<summary>Before</summary>
</details>
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
Adds support for reflecting many more of the input types. This allows those types to be used via scripting, `bevy-inspector-egui`, etc. These types are registered by the `InputPlugin` so that they're automatically available to anyone who wants to use them
Closes#6223
## Solution
Many types now have `#[derive(Reflect, FromReflect)]` added to them in `bevy_input`. Additionally, `#[reflect(traits...)]` has been added for applicable traits to the types.
This PR does not add reflection support for types which have private fields. Notably, `Touch` and `Touches` don't implement `Reflect`/`FromReflect`.
This adds the "glam" feature to the `bevy_reflect` dependency for package `bevy_input`. Since `bevy_input` transitively depends on `glam` already, all this brings in are the reflection `impl`s.
## Migration Guide
- `Input<T>` now implements `Reflect` via `#[reflect]` instead of `#[reflect_value]`. This means it now exposes its private fields via the `Reflect` trait rather than being treated as a value type. For code that relies on the `Input<T>` struct being treated as a value type by reflection, it is still possible to wrap the `Input<T>` type with a wrapper struct and apply `#[reflect_value]` to it.
- As a reminder, private fields exposed via reflection are not subject to any stability guarantees.
---
## Changelog
Added
- Implemented `Reflect` + `FromReflect` for many input-related types. These types are automatically registered when adding the `InputPlugin`.
# Objective
- Proactive changing of code to comply with warnings generated by beta of rustlang version of cargo clippy.
## Solution
- Code changed as recommended by `rustup update`, `rustup default beta`, `cargo run -p ci -- clippy`.
- Tested using `beta` and `stable`. No clippy warnings in either after changes made.
---
## Changelog
- Warnings fixed were: `clippy::explicit-auto-deref` (present in 11 files), `clippy::needless-borrow` (present in 2 files), and `clippy::only-used-in-recursion` (only 1 file).
# Objective
- Some tests are very flaky on a m1
- m1 currently have a 41 ns precision
## Solution
- Do not run tests that compare a `Duration` or a `f64` on a m1 (and m2)
# Objective
- Make the settings of plugins readable during app building
## Solution
- Added a vector of added plugins to the app. Their settings can be accessed as read only
# Objective
- Do not implement `Copy` or `Clone` for `Fetch` types as this is kind of sus soundness wise (it feels like cloning an `IterMut` in safe code to me). Cloning a fetch seems important to think about soundness wise when doing it so I prefer this over adding a `Clone` bound to the assoc type definition (i.e. `type Fetch: Clone`) even though that would also solve the other listed things here.
- Remove a bunch of `QueryFetch<'w, Q>: Clone` bounds from our API as now all fetches can be "cloned" for use in `iter_combinations`. This should also help avoid the type inference regression ptrification introduced where `for<'a> QueryFetch<'a, Q>: Trait` bounds misbehave since we no longer need any of those kind of higher ranked bounds (although in practice we had none anyway).
- Stop being able to "forget" to implement clone for fetches, we've had a lot of issues where either `derive(Clone)` was used instead of a manual impl (so we ended up with too tight bounds on the impl) or flat out forgot to implement Clone at all. With this change all fetches are able to be cloned for `iter_combinations` so this will no longer be possible to mess up.
On an unrelated note, while making this PR I realised we probably want safety invariants on `archetype/table_fetch` that nothing aliases the table_row/archetype_index according to the access we set.
---
## Changelog
`Clone` and `Copy` were removed from all `Fetch` types.
## Migration Guide
- Call `WorldQuery::clone_fetch` instead of `fetch.clone()`. Make sure to add safety comments :)
# Objective
- Build on #6336 for more plugin configurations
## Solution
- `LogSettings`, `ImageSettings` and `DefaultTaskPoolOptions` are now plugins settings rather than resources
---
## Changelog
- `LogSettings` plugin settings have been move to `LogPlugin`, `ImageSettings` to `ImagePlugin` and `DefaultTaskPoolOptions` to `CorePlugin`
## Migration Guide
The `LogSettings` settings have been moved from a resource to `LogPlugin` configuration:
```rust
// Old (Bevy 0.8)
app
.insert_resource(LogSettings {
level: Level::DEBUG,
filter: "wgpu=error,bevy_render=info,bevy_ecs=trace".to_string(),
})
.add_plugins(DefaultPlugins)
// New (Bevy 0.9)
app.add_plugins(DefaultPlugins.set(LogPlugin {
level: Level::DEBUG,
filter: "wgpu=error,bevy_render=info,bevy_ecs=trace".to_string(),
}))
```
The `ImageSettings` settings have been moved from a resource to `ImagePlugin` configuration:
```rust
// Old (Bevy 0.8)
app
.insert_resource(ImageSettings::default_nearest())
.add_plugins(DefaultPlugins)
// New (Bevy 0.9)
app.add_plugins(DefaultPlugins.set(ImagePlugin::default_nearest()))
```
The `DefaultTaskPoolOptions` settings have been moved from a resource to `CorePlugin::task_pool_options`:
```rust
// Old (Bevy 0.8)
app
.insert_resource(DefaultTaskPoolOptions::with_num_threads(4))
.add_plugins(DefaultPlugins)
// New (Bevy 0.9)
app.add_plugins(DefaultPlugins.set(CorePlugin {
task_pool_options: TaskPoolOptions::with_num_threads(4),
}))
```
# Objective
This is a follow-up to #6317, which makes use of a feature of the newest `trybuild` version, `1.0.71`, but does not specify the new patch version in `bevy_ecs_compile_fail_tests/Cargo.toml`.
The PR passed CI because CI downloaded the latest `trybuild` version satisfying the dependency specification. However, Cargo will not know an update is required if a user already has a `^1.0` version of `trybuild` cached locally, which causes the new `$N` syntax to fail the tests.
## Solution
Updated the `trybuild` requirement to `1.0.71`.
# Objective
- Reverts unnecessary version increase for `thiserror` caused by the following PR. 9066d51420
- The aforementioned PR should have increased `thiserrror` version uniformly across all bevy crates. As far as I can tell it was unneccessary to bump versions
## Solution
- Revert versions to the matching version used by other bevy "crates"
```
MBP-Larry-Du.local:~/Code/bevy:$ git grep thiserror
CHANGELOG.md:- [Derive thiserror::Error for HexColorError][2740]
crates/bevy_asset/Cargo.toml:thiserror = "1.0"
crates/bevy_asset/src/asset_server.rs:use thiserror::Error;
crates/bevy_asset/src/io/mod.rs:use thiserror::Error;
crates/bevy_gltf/Cargo.toml:thiserror = "1.0"
crates/bevy_gltf/src/loader.rs:use thiserror::Error;
crates/bevy_input/Cargo.toml:thiserror = "1.0"
crates/bevy_input/src/gamepad.rs:use thiserror::Error;
crates/bevy_reflect/Cargo.toml:thiserror = "1.0"
crates/bevy_reflect/src/path.rs:use thiserror::Error;
crates/bevy_render/Cargo.toml:thiserror = "1.0"
```
---
## Changelog
> This section is optional. If this was a trivial fix, or has no externally-visible impact, you can delete this section.
- What changed as a result of this PR? Fixed dependency conflict for building projects.
Current build of StarRust runs successfully with the `thiserror` reversion: https://github.com/LarsDu/StarRust
But will run into dependency conflicts if `thiserror` is version 1.037
Co-authored-by: Larry Du <larry.du@freenome.com>
# Objective
I was trying to implement a collision system for my game, and believed that the iter_combinations method might be what I need. But I couldn't find a simple explanation of what a combination was in Bevy and thought it could use some more explanation.
## Solution
I added some description to the documentation that can hopefully further elaborate on what a combination is.
I also changed up the docs for the method because a combination is a different thing than a permutation but the Bevy docs seemed to use them interchangeably.
# Objective
- `QueryCombinationIter` can have sizes greater than `usize::MAX`.
- Fixes#5846
## Solution
- Only the implementation of `ExactSizeIterator` has been removed. Instead of using `query_combination.len()`, you can use `query_combination.size_hint().0` to get the same value as before.
---
## Migration Guide
- Switch to using other methods of getting the length.
# Objective
Fixes#5884#2879
Alternative to #2988#5885#2886
"Immutable" Plugin settings are currently represented as normal ECS resources, which are read as part of plugin init. This presents a number of problems:
1. If a user inserts the plugin settings resource after the plugin is initialized, it will be silently ignored (and use the defaults instead)
2. Users can modify the plugin settings resource after the plugin has been initialized. This creates a false sense of control over settings that can no longer be changed.
(1) and (2) are especially problematic and confusing for the `WindowDescriptor` resource, but this is a general problem.
## Solution
Immutable Plugin settings now live on each Plugin struct (ex: `WindowPlugin`). PluginGroups have been reworked to support overriding plugin values. This also removes the need for the `add_plugins_with` api, as the `add_plugins` api can use the builder pattern directly. Settings that can be used at runtime continue to be represented as ECS resources.
Plugins are now configured like this:
```rust
app.add_plugin(AssetPlugin {
watch_for_changes: true,
..default()
})
```
PluginGroups are now configured like this:
```rust
app.add_plugins(DefaultPlugins
.set(AssetPlugin {
watch_for_changes: true,
..default()
})
)
```
This is an alternative to #2988, which is similar. But I personally prefer this solution for a couple of reasons:
* ~~#2988 doesn't solve (1)~~ #2988 does solve (1) and will panic in that case. I was wrong!
* This PR directly ties plugin settings to Plugin types in a 1:1 relationship, rather than a loose "setup resource" <-> plugin coupling (where the setup resource is consumed by the first plugin that uses it).
* I'm not a huge fan of overloading the ECS resource concept and implementation for something that has very different use cases and constraints.
## Changelog
- PluginGroups can now be configured directly using the builder pattern. Individual plugin values can be overridden by using `plugin_group.set(SomePlugin {})`, which enables overriding default plugin values.
- `WindowDescriptor` plugin settings have been moved to `WindowPlugin` and `AssetServerSettings` have been moved to `AssetPlugin`
- `app.add_plugins_with` has been replaced by using `add_plugins` with the builder pattern.
## Migration Guide
The `WindowDescriptor` settings have been moved from a resource to `WindowPlugin::window`:
```rust
// Old (Bevy 0.8)
app
.insert_resource(WindowDescriptor {
width: 400.0,
..default()
})
.add_plugins(DefaultPlugins)
// New (Bevy 0.9)
app.add_plugins(DefaultPlugins.set(WindowPlugin {
window: WindowDescriptor {
width: 400.0,
..default()
},
..default()
}))
```
The `AssetServerSettings` resource has been removed in favor of direct `AssetPlugin` configuration:
```rust
// Old (Bevy 0.8)
app
.insert_resource(AssetServerSettings {
watch_for_changes: true,
..default()
})
.add_plugins(DefaultPlugins)
// New (Bevy 0.9)
app.add_plugins(DefaultPlugins.set(AssetPlugin {
watch_for_changes: true,
..default()
}))
```
`add_plugins_with` has been replaced by `add_plugins` in combination with the builder pattern:
```rust
// Old (Bevy 0.8)
app.add_plugins_with(DefaultPlugins, |group| group.disable::<AssetPlugin>());
// New (Bevy 0.9)
app.add_plugins(DefaultPlugins.build().disable::<AssetPlugin>());
```
# Objective
Scenes are currently represented as a list of entities. This is all we need currently, but we may want to add more data to this format in the future (metadata, asset lists, etc.).
It would be nice to update the format in preparation of possible future changes. Doing so now (i.e., before 0.9) could mean reduced[^1] breakage for things added in 0.10.
[^1]: Obviously, adding features runs the risk of breaking things regardless. But if all features added are for whatever reason optional or well-contained, then users should at least have an easier time updating.
## Solution
Made the scene root a struct rather than a list.
```rust
(
entities: [
// Entity data here...
]
)
```
---
## Changelog
* The scene format now puts the entity list in a newly added `entities` field, rather than having it be the root object
## Migration Guide
The scene file format now uses a struct as the root object rather than a list of entities. The list of entities is now found in the `entities` field of this struct.
```rust
// OLD
[
(
entity: 0,
components: [
// Components...
]
),
]
// NEW
(
entities: [
(
entity: 0,
components: [
// Components...
]
),
]
)
```
Co-authored-by: Gino Valente <49806985+MrGVSV@users.noreply.github.com>
# Objective
- You usually want to say that a given animation *should* be playing, doing nothing if it's already playing.
## Solution
- Rename play to start and add new play method that won't overwrite the existing animation if it's already playing #6350
---
## Changelog
### Changed
`AnimationPlayer::play` will now not restart the animation if it's already playing
### Added
An `AnimationPlayer ::start` method, which has the old behavior of `play`
## Migration guide
- If you were using `play` to restart an animation that was already playing, that functionality has been moved to `start`. Now, `play` won't have any effect if the requested animation is already playing.
# Objective
Improve ergonomics by passing on the `IntoIterator` impl of the underlying type to wrapper types.
## Solution
Implement `IntoIterator` for ECS wrapper types (Mut, Local, Res, etc.).
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
- Improve #3953
## Solution
- The very specific circumstances under which the render world is reset meant that the flush_as_invalid function could be replaced with one that had a noop as its init method.
- This removes a double-writing issue leading to greatly increased performance.
Running the reproduction code in the linked issue, this change nearly doubles the framerate.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
- Avoids creating a `SurfaceConfiguration` for every window in every frame for the `prepare_windows` system
- As such also avoid calling `get_supported_formats` for every window in every frame
## Solution
- Construct `SurfaceConfiguration` lazyly in `prepare_windows`
---
This also changes the error message for failed initial surface configuration from "Failed to acquire next swapchain texture" to "Error configuring surface".
# Objective
When running the scene example, you might notice we end up printing out the following:
```ron
// ...
{
"scene::ComponentB": (
value: "hello",
_time_since_startup: (
secs: 0,
nanos: 0,
),
),
},
// ...
```
We should not be printing out `_time_since_startup` as the field is marked with `#[reflect(skip_serializing)]`:
```rust
#[derive(Component, Reflect)]
#[reflect(Component)]
struct ComponentB {
pub value: String,
#[reflect(skip_serializing)]
pub _time_since_startup: Duration,
}
```
This is because when we create the `DynamicScene`, we end up calling `Reflect::clone_value`:
82126697ee/crates/bevy_scene/src/dynamic_scene_builder.rs (L114-L114)
This results in non-Value types being cloned into Dynamic types, which means the `TypeId` returned from `reflected_value.type_id()` is not the same as the original component's.
And this meant we were not able to locate the correct `TypeRegistration`.
## Solution
Use `TypeInfo::type_id()` instead of calling `Any::type_id()` on the value directly.
---
## Changelog
* Fix a bug introduced in `0.9.0-dev` where scenes disregarded component's type registrations
# Objective
- Clipping (visible in the UI example with text scrolling) is funky
- Fixes#6287
## Solution
- Fix UV calculation:
- correct order for values (issue introduced in #6000)
- add the `y` values instead of subtracting them now that vertical order is reversed
- take scale factor into account (bug already present before reversing the order)
- While around clipping, I changed clip to only mutate when changed
No more funkiness! 😞
<img width="696" alt="Screenshot 2022-10-23 at 22 44 18" src="https://user-images.githubusercontent.com/8672791/197417721-30ad4150-5264-427f-ac82-e5265c1fb3a9.png">
# Objective
Fixes#6339.
## Solution
This PR adds a new type, `GamepadInfo`, which holds metadata associated with a particular `Gamepad`. The `Gamepads` resource now holds a `HashMap<Gamepad, GamepadInfo>`. The `GamepadInfo` is created when the gamepad backend (by default `bevy_gilrs`) emits a "gamepad connected" event.
The `gamepad_viewer` example has been updated to showcase the new functionality.
Before:
After:
---
## Changelog
### Added
- Added `GamepadInfo`.
- Added `Gamepads::name()`, which returns the name of the specified gamepad if it exists.
### Changed
- `GamepadEventType::Connected` is now a tuple variant with a single field of type `GamepadInfo`.
- Since `GamepadInfo` is not `Copy`, `GamepadEventType` is no longer `Copy`. The same is true of `GamepadEvent` and `GamepadEventRaw`.
## Migration Guide
- Pattern matches on `GamepadEventType::Connected` will need to be updated, as the form of the variant has changed.
- Code that requires `GamepadEvent`, `GamepadEventRaw` or `GamepadEventType` to be `Copy` will need to be updated.
I found myself doing
```rust
let child = commands.spawn(..).id();
commands.entity(parent).add_child(child);
```
When that could just be
```rust
commands.spawn(..).set_parent(parent);
```
Adding `set_parent` was trivial as it's just an `AddChild` command. Most of the changes are for `remove_parent`.
Also updated some outdated docs.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
Adds a better interface for performing mathematical operations with UI unit `Val`. Fixes#6080.
## Solution
- Added `try_add` and `try_sub` methods to Val.
- Removed the `Add` and `AddAssign` impls for `Val` that introduced unintuitive and bug-prone behaviour.
- As a consequence of the prior, ~~changed the `Add` and `Sub` impls for the `Size` struct to take a `(Val, Val)` instead of `Vec2`~~ deleted the `Add` and `Sub` impls for the `Size` struct
- Added a `From<(Val, Val)>` impl for the `Size` struct
- Added `evaluate(size: f32)` method that converts from `Val::Percent` to `Val::Px`.
- Added `try_add_with_size` and `try_sub_with_size` methods to `Val`, which evaluate `Val::Percent` values into `Val::Px` values before adding.
---
## Migration Guide
Instead of using the + and - operators, perform calculations on `Val`s using the new `try_add` and `try_sub` methods. Multiplication and division remained unchanged. Also, when adding or subtracting from `Size`, ~~use a `Val` tuple instead of `Vec2`~~ perform the addition on `width` and `height` separately.
Co-authored-by: Dawid Piotrowski <41804418+Pietrek14@users.noreply.github.com>
# Objective
Fixes#5559
Replaces #5628
## Solution
Because the generated method from_components() creates an instance of Self my implementation requires any field type that is marked to be ignored to implement Default.
---
## Changelog
Added the possibility to ignore fields in a bundle with `#[bundle(ignore)]`. Typically used when `PhantomData` needs to be added to a `Bundle`.
# Objective
- #4466 broke local tasks running.
- Fixes https://github.com/bevyengine/bevy/issues/6120
## Solution
- Add system for ticking local executors on main thread into bevy_core where the tasks pools are initialized.
- Add ticking local executors into thread executors
## Changelog
- tick all thread local executors in task pool.
## Notes
- ~~Not 100% sure about this PR. Ticking the local executor for the main thread in scope feels a little kludgy as it requires users of bevy_tasks to be calling scope periodically for those tasks to make progress.~~ took this out in favor of a system that ticks the local executors.
# Objective
- Fix disabling features in bevy_ecs (broken by #5630)
- Add tests in CI for bevy_ecs, bevy_reflect and bevy as those crates could be use standalone
Add the following message:
```
Items are returned in the order of the list of entities.
Entities that don't match the query are skipped.
```
Additionally, the docs in `iter.rs` and `state.rs` were updated to match those in `query.rs`.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
- Add Time-Adjusted Rolling EMA-based smoothing to diagnostics.
- Closes#4983; see that issue for more more information.
## Terms
- EMA - [Exponential Moving Average](https://en.wikipedia.org/wiki/Moving_average#Exponential_moving_average)
- SMA - [Simple Moving Average](https://en.wikipedia.org/wiki/Moving_average#Simple_moving_average)
## Solution
- We use a fairly standard approximation of a true EMA where $EMA_{\text{frame}} = EMA_{\text{previous}} + \alpha \left( x_{\text{frame}} - EMA_{\text{previous}} \right)$ where $\alpha = \Delta t / \tau$ and $\tau$ is an arbitrary smoothness factor. (See #4983 for more discussion of the math.)
- The smoothness factor is here defaulted to $2 / 21$; this was chosen fairly arbitrarily as supposedly related to the existing 20-bucket SMA.
- The smoothness factor can be set on a per-diagnostic basis via `Diagnostic::with_smoothing_factor`.
---
## Changelog
### Added
- `Diagnostic::smoothed` - provides an exponentially smoothed view of a recorded diagnostic, to e.g. reduce jitter in frametime readings.
### Changed
- `LogDiagnosticsPlugin` now records the smoothed value rather than the raw value.
- For diagnostics recorded less often than every 0.1 seconds, this change to defaults will have no visible effect.
- For discrete diagnostics where this smoothing is not desirable, set a smoothing factor of 0 to disable smoothing.
- The average of the recent history is still shown when available.
# Objective
At least partially addresses #6282.
Resources are currently stored as a dedicated Resource archetype (ID 1). This allows for easy code reusability, but unnecessarily adds 72 bytes (on 64-bit systems) to the struct that is only used for that one archetype. It also requires several fields to be `pub(crate)` which isn't ideal.
This should also remove one sparse-set lookup from fetching, inserting, and removing resources from a `World`.
## Solution
- Add `Resources` parallel to `Tables` and `SparseSets` and extract the functionality used by `Archetype` in it.
- Remove `unique_components` from `Archetype`
- Remove the `pub(crate)` on `Archetype::components`.
- Remove `ArchetypeId::RESOURCE`
- Remove `Archetypes::resource` and `Archetypes::resource_mut`
---
## Changelog
Added: `Resources` type to store resources.
Added: `Storages::resource`
Removed: `ArchetypeId::RESOURCE`
Removed: `Archetypes::resource` and `Archetypes::resources`
Removed: `Archetype::unique_components` and `Archetypes::unique_components_mut`
## Migration Guide
Resources have been moved to `Resources` under `Storages` in `World`. All code dependent on `Archetype::unique_components(_mut)` should access it via `world.storages().resources()` instead.
All APIs accessing the raw data of individual resources (mutable *and* read-only) have been removed as these APIs allowed for unsound unsafe code. All usages of these APIs should be changed to use `World::{get, insert, remove}_resource`.
# Objective
Speed up queries that are fragmented over many empty archetypes and tables.
## Solution
Add a early-out to check if the table or archetype is empty before iterating over it. This adds an extra branch for every archetype matched, but skips setting the archetype/table to the underlying state and any iteration over it.
This may not be worth it for the default `Query::iter` and maybe even the `Query::for_each` implementations, but this definitely avoids scheduling unnecessary tasks in the `Query::par_for_each` case.
Ideally, `matched_archetypes` should only contain archetypes where there's actually work to do, but this would add a `O(n)` flat cost to every call to `update_archetypes` that scales with the number of matched archetypes.
TODO: Benchmark
# Objective
- Make `Time` API more consistent.
- Support time accel/decel/pause.
## Solution
This is just the `Time` half of #3002. I was told that part isn't controversial.
- Give the "delta time" and "total elapsed time" methods `f32`, `f64`, and `Duration` variants with consistent naming.
- Implement accelerating / decelerating the passage of time.
- Implement stopping time.
---
## Changelog
- Changed `time_since_startup` to `elapsed` because `time.time_*` is just silly.
- Added `relative_speed` and `set_relative_speed` methods.
- Added `is_paused`, `pause`, `unpause` , and methods. (I'd prefer `resume`, but `unpause` matches `Timer` API.)
- Added `raw_*` variants of the "delta time" and "total elapsed time" methods.
- Added `first_update` method because there's a non-zero duration between startup and the first update.
## Migration Guide
- `time.time_since_startup()` -> `time.elapsed()`
- `time.seconds_since_startup()` -> `time.elapsed_seconds_f64()`
- `time.seconds_since_startup_wrapped_f32()` -> `time.elapsed_seconds_wrapped()`
If you aren't sure which to use, most systems should continue to use "scaled" time (e.g. `time.delta_seconds()`). The realtime "unscaled" time measurements (e.g. `time.raw_delta_seconds()`) are mostly for debugging and profiling.
# Objective
Use saturate wgsl function now implemented in naga (version 0.10.0). There is now no need for one in utils.wgsl.
naga's version allows usage for not only scalars but vectors as well.
## Solution
Remove the utils.wgsl saturate function.
## Changelog
Remove saturate function from utils.wgsl in favor of saturate in naga v0.10.0.
# Objective
I recently wanted to look at the possibility of adding `Mutated` and `Unchanged` query filters and was confronted with some seemingly unrelated broken tests.
These tests were written in such a way that changing the number of WorldQuery impls in the project would break them.
Fortunately, a [very recent release of trybuild](https://github.com/dtolnay/trybuild/releases/tag/1.0.70) has made this unnecessary.
## Solution
Replace hardcoded numbers in test output with `$N` placeholders.
# Objective
Fixes https://github.com/bevyengine/bevy/issues/6306
## Solution
Change the failing assert and expand example to explain when ordering is deterministic or not.
Co-authored-by: Mike Hsu <mike.hsu@gmail.com>
# Objective
The `RenderLayers` type is never registered, making it unavailable for reflection.
## Solution
Register it in `CameraPlugin`, the same plugin that registers the related `Visibility*` types.
# Objective
- Update `wgpu` to 0.14.0, `naga` to `0.10.0`, `winit` to 0.27.4, `raw-window-handle` to 0.5.0, `ndk` to 0.7.
## Solution
---
## Changelog
### Changed
- Changed `RawWindowHandleWrapper` to `RawHandleWrapper` which wraps both `RawWindowHandle` and `RawDisplayHandle`, which satisfies the `impl HasRawWindowHandle and HasRawDisplayHandle` that `wgpu` 0.14.0 requires.
- Changed `bevy_window::WindowDescriptor`'s `cursor_locked` to `cursor_grab_mode`, change its type from `bool` to `bevy_window::CursorGrabMode`.
## Migration Guide
- Adjust usage of `bevy_window::WindowDescriptor`'s `cursor_locked` to `cursor_grab_mode`, and adjust its type from `bool` to `bevy_window::CursorGrabMode`.
# Objective
Resolves#6197
Make it so that doc comments can be retrieved via reflection.
## Solution
Adds the new `documentation` feature to `bevy_reflect` (disabled by default).
When enabled, documentation can be found using `TypeInfo::doc` for reflected types:
```rust
/// Some struct.
///
/// # Example
///
/// ```ignore
/// let some_struct = SomeStruct;
/// ```
#[derive(Reflect)]
struct SomeStruct;
let info = <SomeStruct as Typed>::type_info();
assert_eq!(
Some(" Some struct.\n\n # Example\n\n ```ignore\n let some_struct = SomeStruct;\n ```"),
info.docs()
);
```
### Notes for Reviewers
The bulk of the files simply added the same 16 lines of code (with slightly different documentation). Most of the real changes occur in the `bevy_reflect_derive` files as well as in the added tests.
---
## Changelog
* Added `documentation` feature to `bevy_reflect`
* Added `TypeInfo::docs` method (and similar methods for all info types)
# Objective
Fixes#6272
## Solution
Revert to old way of positioning text for Text2D rendered text.
Co-authored-by: Michel van der Hulst <hulstmichel@gmail.com>
# Objective
Make toggling the visibility of an entity slightly more convenient.
## Solution
Add a mutating `toggle` method to the `Visibility` component
```rust
fn my_system(mut query: Query<&mut Visibility, With<SomeMarker>>) {
let mut visibility = query.single_mut();
// before:
visibility.is_visible = !visibility.is_visible;
// after:
visibility.toggle();
}
```
## Changelog
### Added
- Added a mutating `toggle` method to the `Visibility` component
# Objective
Fixes#6244
## Solution
Uses derive to implement `Serialize` and `Deserialize` for `Timer` and `Stopwatch`
### Things to consider
- Should fields such as `finished` and `times_finished_this_tick` in `Timer` be serialized?
- Does `Countdown` and `PrintOnCompletionTimer` need to be serialized and deserialized?
## Changelog
Added `Serialize` and `Deserialize` implementations to `Timer` and `Stopwatch`, `Countdown`.
# Objective
Currently, surprising behavior happens when specifying `#[reflect(...)]` or `#[reflect_value(...)]` multiple times. Rather than merging the traits lists from all attributes, only the trait list from the last attribute is used. For example, in the following code, only the `Debug` and `Hash` traits are reflected and not `Default` or `PartialEq`:
```rs
#[derive(Debug, PartialEq, Hash, Default, Reflect)]
#[reflect(PartialEq, Default)]
#[reflect(Debug, Hash)]
struct Foo;
```
This is especially important when some traits should only be reflected under certain circumstances. For example, this previously had surprisingly behavior when the "serialize" feature is enabled:
```rs
#[derive(Debug, Hash, Reflect)]
#[reflect(Debug, Hash)]
#[cfg_attr(
feature = "serialize",
derive(Serialize, Deserialize),
reflect(Serialize, Deserialize)
]
struct Foo;
```
In addition, compile error messages generated from using the derive macro often point to the `#[derive(Reflect)]` rather than to the source of the error. It would be a lot more helpful if the compiler errors pointed to what specifically caused the error rather than just to the derive macro itself.
## Solution
Merge the trait lists in all `#[reflect(...)]` and `#[reflect_value(...)]` attributes. Additionally, make `#[reflect]` and `#[reflect_value]` mutually exclusive.
Additionally, span information is carried throughout some parts of the code now to ensure that error messages point to more useful places and better indicate what caused those errors. For example, `#[reflect(Hash, Hash)]` points to the second `Hash` as the source of an error. Also, in the following example, the compiler error now points to the `Hash` in `#[reflect(Hash)]` rather than to the derive macro:
```rs
#[derive(Reflect)]
#[reflect(Hash)] // <-- compiler error points to `Hash` for lack of a `Hash` implementation
struct Foo;
```
---
## Changelog
Changed
- Using multiple `#[reflect(...)]` or `#[reflect_value(...)]` attributes now merges the trait lists. For example, `#[reflect(Debug, Hash)] #[reflect(PartialEq, Default)]` is equivalent to `#[reflect(Debug, Hash, PartialEq, Default)]`.
- Multiple `#[reflect(...)]` and `#[reflect_value(...)]` attributes were previously accepted, but only the last attribute was respected.
- Using both `#[reflect(...)]` and `#[reflect_value(...)]` was previously accepted, but had surprising behavior. This is no longer accepted.
- Improved error messages for `#[derive(Reflect)]` by propagating useful span information. Many errors should now point to the source of those errors rather than to the derive macro.
# Objective
Fixes https://github.com/bevyengine/bevy/issues/3418
## Solution
Originally a rebase of https://github.com/bevyengine/bevy/pull/3446. Work was originally done by mfdorst, who should receive considerable credit. Then the error types were extensively reworked by targrub.
## Migration Guide
`AxisSettings` now has a `new()`, which may return an `AxisSettingsError`.
`AxisSettings` fields made private; now must be accessed through getters and setters. There's a dead zone, from `.deadzone_upperbound()` to `.deadzone_lowerbound()`, and a live zone, from `.deadzone_upperbound()` to `.livezone_upperbound()` and from `.deadzone_lowerbound()` to `.livezone_lowerbound()`.
`AxisSettings` setters no longer panic.
`ButtonSettings` fields made private; now must be accessed through getters and setters.
`ButtonSettings` now has a `new()`, which may return a `ButtonSettingsError`.
Co-authored-by: targrub <62773321+targrub@users.noreply.github.com>
# Objective
- Trying to make it possible to do write tests that don't require a raw window handle.
- Fixes https://github.com/bevyengine/bevy/issues/6106.
## Solution
- Make the interface and type changes. Avoid accessing `None`.
---
## Changelog
- Converted `raw_window_handle` field in both `Window` and `ExtractedWindow` to `Option<RawWindowHandleWrapper>`.
- Revised accessor function `Window::raw_window_handle()` to return `Option<RawWindowHandleWrapper>`.
- Skip conditions in loops that would require a raw window handle (to create a `Surface`, for example).
## Migration Guide
`Window::raw_window_handle()` now returns `Option<RawWindowHandleWrapper>`.
Co-authored-by: targrub <62773321+targrub@users.noreply.github.com>
As suggested in #6104, it would be nice to link directly to `linux_dependencies.md` file in the panic message when running on Linux. And when not compiling for Linux, we fall back to the old message.
Signed-off-by: Lena Milizé <me@lvmn.org>
# Objective
Resolves#6104.
## Solution
Add link to `linux_dependencies.md` when compiling for Linux, and fall back to the old one when not.
# Objective
- Fixes#6206
## Solution
- Create a constructor for creating `ReflectComponent` and `ReflectResource`
---
## Changelog
> This section is optional. If this was a trivial fix, or has no externally-visible impact, you can delete this section.
### Added
- Created constructors for `ReflectComponent` and `ReflectResource`, allowing for advanced scripting use-cases.
# Objective
There is currently no good way of getting the width (# of components) of a table outside of `bevy_ecs`.
# Solution
Added the methods `Table::{component_count, component_capacity}`
For consistency and clarity, renamed `Table::{len, capacity}` to `entity_count` and `entity_capacity`.
## Changelog
- Added the methods `Table::component_count` and `Table::component_capacity`
- Renamed `Table::len` and `Table::capacity` to `entity_count` and `entity_capacity`
## Migration Guide
Any use of `Table::len` should now be `Table::entity_count`. Any use of `Table::capacity` should now be `Table::entity_capacity`.
As mentioned in #2926, it's better to have an explicit type that clearly communicates the intent of the timer mode rather than an opaque boolean, which can be only understood when knowing the signature or having to look up the documentation.
This also opens up a way to merge different timers, such as `Stopwatch`, and possibly future ones, such as `DiscreteStopwatch` and `DiscreteTimer` from #2683, into one struct.
Signed-off-by: Lena Milizé <me@lvmn.org>
# Objective
Fixes#2926.
## Solution
Introduce `TimerMode` which replaces the `bool` argument of `Timer` constructors. A `Default` value for `TimerMode` is `Once`.
---
## Changelog
### Added
- `TimerMode` enum, along with variants `TimerMode::Once` and `TimerMode::Repeating`
### Changed
- Replace `bool` argument of `Timer::new` and `Timer::from_seconds` with `TimerMode`
- Change `repeating: bool` field of `Timer` with `mode: TimerMode`
## Migration Guide
- Replace `Timer::new(duration, false)` with `Timer::new(duration, TimerMode::Once)`.
- Replace `Timer::new(duration, true)` with `Timer::new(duration, TimerMode::Repeating)`.
- Replace `Timer::from_seconds(seconds, false)` with `Timer::from_seconds(seconds, TimerMode::Once)`.
- Replace `Timer::from_seconds(seconds, true)` with `Timer::from_seconds(seconds, TimerMode::Repeating)`.
- Change `timer.repeating()` to `timer.mode() == TimerMode::Repeating`.
# Objective
- Add a way to iterate over all entities from &World
## Solution
- Added a function `iter_entities` on World which returns an iterator of `Entity` derived from the entities in the `World`'s `archetypes`
---
## Changelog
- Added a function `iter_entities` on World, allowing iterating over all entities in contexts where you only have read-only access to the World.
# Objective
Fixes#5820
## Solution
Change field name and documentation from `bevy::ui::Node` struct
---
## Changelog
`bevy::ui::Node` `size` field has renamed to `calculated_size`
## Migration Guide
All references to the old `size` name has been changed, to access `bevy::ui::Node` `size` field use `calculated_size`
Bevy's coordinate system is right-handed Y up, so +Z points towards my nose and I'm looking in the -Z direction. Therefore, `Transform::looking_at/look_at` must be pointing towards -Z. Or am I wrong here?
# Objective
- make it easier to build dynamic scenes
## Solution
- add a builder to create a dynamic scene from a world. it can extract an entity or an iterator of entities
- alternative to #6013, leaving the "hierarchy iteration" part to #6185 which does it better
- alternative to #6004
- using a builder makes it easier to chain several extractions
# Objective
> System chaining is a confusing name: it implies the ability to construct non-linear graphs, and suggests a sense of system ordering that is only incidentally true. Instead, it actually works by passing data from one system to the next, much like the pipe operator.
> In the accepted [stageless RFC](https://github.com/bevyengine/rfcs/blob/main/rfcs/45-stageless.md), this concept is renamed to piping, and "system chaining" is used to construct groups of systems with ordering dependencies between them.
Fixes#6225.
## Changelog
System chaining has been renamed to system piping to improve clarity (and free up the name for new ordering APIs).
## Migration Guide
The `.chain(handler_system)` method on systems is now `.pipe(handler_system)`.
The `IntoChainSystem` trait is now `IntoPipeSystem`, and the `ChainSystem` struct is now `PipeSystem`.
# Objective
`RemoveChildren` could remove the `Parent` component from children belonging to a different parent, which breaks the hierarchy.
This change looks a little funny because I'm reusing the events to avoid needing to clone the parent's `Children`.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
- Taking the API improvement out of #5431
- `iter_instance_entities` used to return an option of iterator, now it just returns an iterator
---
## Changelog
- If you use `SceneSpawner::iter_instance_entities`, it no longer returns an `Option`. The iterator will be empty if the return value used to be `None`
# Objective
- Adding Debug implementations for App, Stage, Schedule, Query, QueryState.
- Fixes#1130.
## Solution
- Implemented std::fmt::Debug for a number of structures.
---
## Changelog
Also added Debug implementations for ParallelSystemExecutor, SingleThreadedExecutor, various RunCriteria structures, SystemContainer, and SystemDescriptor.
Opinions are sure to differ as to what information to provide in a Debug implementation. Best guess was taken for this initial version for these structures.
Co-authored-by: targrub <62773321+targrub@users.noreply.github.com>
See commit message.
I noticed I couldn't use `globals.time` when using `Material2d`.
I copied the solution from 8073362039 , and now `Material2d` works for me.
Perhaps some of these struct definitions could be shared in the future, but for now I've just copy pasted it (it looked like the `View` struct was done that way).
Ping @IceSentry , I saw a comment on the linked commit that you intended to do this work at some point in the future.
# Objective
- It's possible to create a mesh without positions or normals, but currently bevy forces these attributes to be present on any mesh.
## Solution
- Don't assume these attributes are present and add a shader defs for each attributes
- I updated 2d and 3d meshes to use the same logic.
---
## Changelog
- Meshes don't require any attributes
# Notes
I didn't update the pbr.wgsl shader because I'm not sure how to handle it. It doesn't really make sense to use it without positions or normals.
# Objective
- `process_touch_event` in `bevy_input` don't update position info. `TouchPhase::Ended` and `TouchPhase::Cancelled` should use the position info from `pressed`. Otherwise, it'll not update. The position info is updated from `TouchPhase::Moved`.
## Solution
- Use updated touch info.
---
## Changelog
> This section is optional. If this was a trivial fix, or has no externally-visible impact, feel free to skip this section.
- Fixed: bevy_input, fix process touch event, update touch info
# Objective
When designing an API, you may wish to provide access only to a specific field of a component or resource. The current options for doing this in safe code are
* `*Mut::into_inner`, which flags a change no matter what.
* `*Mut::bypass_change_detection`, which misses all changes.
## Solution
Add the method `map_unchanged`.
### Example
```rust
// When run, zeroes the translation of every entity.
fn reset_all(mut transforms: Query<&mut Transform>) {
for transform in &mut transforms {
// We pinky promise not to modify `t` within the closure.
let translation = transform.map_unchanged(|t| &mut t.translation);
// Only reset the translation if it isn't already zero.
translation.set_if_not_equal(Vec2::ZERO);
}
}
```
---
## Changelog
+ Added the method `map_unchanged` to types `Mut<T>`, `ResMut<T>`, and `NonSendMut<T>`.
This is a holdover from back when `Transform` was backed by a private `Mat4` two years ago.
Not particularly useful anymore :)
## Migration Guide
`Transform::apply_non_uniform_scale` has been removed.
It can be replaced with the following snippet:
```rust
transform.scale *= scale_factor;
```
Co-authored-by: devil-ira <justthecooldude@gmail.com>
The docs ended up quite verbose :v
Also added a missing `#[inline]` to `GlobalTransform::mul_transform`.
I'd say this resolves#5500
# Migration Guide
`Transform::mul_vec3` has been renamed to `transform_point`.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
Make `GlobalTransform` constructible from scripts, in the same vein as #6187.
## Solution
- Use the derive macro to reflect default
---
## Changelog
> This section is optional. If this was a trivial fix, or has no externally-visible impact, you can delete this section.
- `GlobalTransform` now reflects the `Default` trait.
…
# Objective
- Fixes Camera not being serializable due to missing registrations in core functionality.
- Fixes#6169
## Solution
- Updated Bevy_Render CameraPlugin with registrations for Option<Viewport> and then Bevy_Core CorePlugin with registrations for ReflectSerialize and ReflectDeserialize for type data Range<f32> respectively according to the solution in #6169
Co-authored-by: Noah <noahshomette@gmail.com>
# Background
Incremental implementation of #4299. The code is heavily borrowed from that PR.
# Objective
The execution order ambiguity checker often emits false positives, since bevy is not aware of invariants upheld by the user.
## Solution
Title
---
## Changelog
+ Added methods `SystemDescriptor::ignore_all_ambiguities` and `::ambiguous_with`. These allow you to silence warnings for specific system-order ambiguities.
## Migration Guide
***Note for maintainers**: This should replace the migration guide for #5916*
Ambiguity sets have been replaced with a simpler API.
```rust
// These systems technically conflict, but we don't care which order they run in.
fn jump_on_click(mouse: Res<Input<MouseButton>>, mut transforms: Query<&mut Transform>) { ... }
fn jump_on_spacebar(keys: Res<Input<KeyCode>>, mut transforms: Query<&mut Transform>) { ... }
//
// Before
#[derive(AmbiguitySetLabel)]
struct JumpSystems;
app
.add_system(jump_on_click.in_ambiguity_set(JumpSystems))
.add_system(jump_on_spacebar.in_ambiguity_set(JumpSystems));
//
// After
app
.add_system(jump_on_click.ambiguous_with(jump_on_spacebar))
.add_system(jump_on_spacebar);
```
# Objective
There is no Srgb support on some GPU and display protocols with `winit` (for example, Nvidia's GPUs with Wayland). Thus `TextureFormat::bevy_default()` which returns `Rgba8UnormSrgb` or `Bgra8UnormSrgb` will cause panics on such platforms. This patch will resolve this problem. Fix https://github.com/bevyengine/bevy/issues/3897.
## Solution
Make `initialize_renderer` expose `wgpu::Adapter` and `first_available_texture_format`, use the `first_available_texture_format` by default.
## Changelog
* Fixed https://github.com/bevyengine/bevy/issues/3897.
# Objective
Closes#6202.
The default background color for `NodeBundle` is currently white.
However, it's very rare that you actually want a white background color.
Instead, you often want a background color specific to the style of your game or a transparent background (e.g. for UI layout nodes).
## Solution
`Default` is not derived for `NodeBundle` anymore, but explicitly specified.
The default background color is now transparent (`Color::NONE.into()`) as this is the most common use-case, is familiar from the web and makes specifying a layout for your UI less tedious.
---
## Changelog
- Changed the default `NodeBundle.background_color` to be transparent (`Color::NONE.into()`).
## Migration Guide
If you want a `NodeBundle` with a white background color, you must explicitly specify it:
Before:
```rust
let node = NodeBundle {
..default()
}
```
After:
```rust
let node = NodeBundle {
background_color: Color::WHITE.into(),
..default()
}
```
# Objective
Allow `Mesh2d` shaders to work with meshes that have vertex tangents
## Solution
Correctly pass `mesh.model` into `mesh2d_tangent_local_to_world`
# Objective
Fixes#6010
## Solution
As discussed in #6010, this makes it so the `Children` component is removed from the entity whenever all of its children are removed. The behavior is now consistent between all of the commands that may remove children from a parent, and this is tested via two new test functions (one for world functions and one for commands).
Documentation was also added to `insert_children`, `push_children`, `add_child` and `remove_children` commands to make this behavior clearer for users.
## Changelog
- Fixed `Children` component not getting removed from entity when all its children are moved to a new parent.
## Migration Guide
- Queries with `Changed<Children>` will no longer match entities that had all of their children removed using `remove_children`.
- `RemovedComponents<Children>` will now contain entities that had all of their children remove using `remove_children`.
# Objective
- Reflecting `Default` is required for scripts to create `Reflect` types at runtime with no static type information.
- Reflecting `Default` on `Handle<T>` and `ComputedVisibility` should allow scripts from `bevy_mod_js_scripting` to actually spawn sprites from scratch, without needing any hand-holding from the host-game.
## Solution
- Derive `ReflectDefault` for `Handle<T>` and `ComputedVisiblity`.
---
## Changelog
> This section is optional. If this was a trivial fix, or has no externally-visible impact, you can delete this section.
- The `Default` trait is now reflected for `Handle<T>` and `ComputedVisibility`
# Objective
- Field `id` of `Handle<T>` is public: https://docs.rs/bevy/latest/bevy/asset/struct.Handle.html#structfield.id
- Changing the value of this field doesn't make sense as it could mean changing the previous handle without dropping it, breaking asset cleanup detection for the old handle and the new one
## Solution
- Make the field private, and add a public getter
Opened after discussion in #6171. Pinging @zicklag
---
## Migration Guide
- If you were accessing the value `handle.id`, you can now do so with `handle.id()`
# Objective
Add a method for getting a world space ray from a viewport position.
Opted to add a `Ray` type to `bevy_math` instead of returning a tuple of `Vec3`'s as this is clearer and easier to document
The docs on `viewport_to_world` are okay, but I'm not super happy with them.
## Changelog
* Add `Camera::viewport_to_world`
* Add `Camera::ndc_to_world`
* Add `Ray` to `bevy_math`
* Some doc tweaks
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
- Fix#5285
## Solution
- Put the panicking system in a single threaded stage during the test
- This way only the main thread will panic, which is handled by `cargo test`
# Objective
- Fixes contradictory docs in Window::PresentMode partaining to PresentMode fallback behavior. Fix based on commit history showing the most recent update didn't remove old references to the gracefal fallback for Immediate and Mailbox.
- Fixes#5831
## Solution
- Updated the docs for Window::PresentMode itself and for each individual enum variant to clarify which will fallback and which will panic.
Co-authored-by: Noah <noahshomette@gmail.com>
# Objective
fix error with pbr shader's spotlight direction calculation when direction.y ~= 0
## Solution
in pbr_lighting.wgsl, clamp `1-x^2-z^2` to `>= 0` so that we can safely `sqrt` it
# Objective
The `Wireframe` type implements `Reflect`, but is never registered, making its reflection inaccessible.
## Solution
Call `App::register_type::<Wireframe>()` in the `Plugin::build` implementation of `WireframePlugin`.
---
## Changelog
Fixed `Wireframe` type reflection not getting registered.
# Objective
Add more documentation on `StandardMaterial` and improve
consistency on existing doc.
Co-authored-by: Nicola Papale <nicopap@users.noreply.github.com>
# Objective
Relaxes the trait bound for `World::resource_scope` to allow non-send resources. Fixes#6037.
## Solution
No big changes in code had to be made. Added a check so that the non-send resources won't be accessed from a different thread.
---
## Changelog
- `World::resource_scope` accepts non-send resources now
- `World::resource_scope` verifies non-send access if the resource is non-send
- Two new tests are added, one for valid use of `World::resource_scope` with a non-send resource, and one for invalid use (calling it from a different thread, resulting in panic)
Co-authored-by: Dawid Piotrowski <41804418+Pietrek14@users.noreply.github.com>
# Objective
As explained by #5960, `Commands::get_or_spawn` may return a dangling `EntityCommands` that references a non-existing entities. As explained in [this comment], it may be undesirable to make the method return an `Option`.
- Addresses #5960
- Alternative to #5961
## Solution
This PR adds a doc comment to the method to inform the user that the returned `EntityCommands` is not guaranteed to be valid. It also adds panic doc comments on appropriate `EntityCommands` methods.
[this comment]: https://github.com/bevyengine/bevy/pull/5961#issuecomment-1259870849
# Objective
- Alpha mask was previously ignored when using an unlit material.
- Fixes https://github.com/bevyengine/bevy/issues/4479
## Solution
- Extract the alpha discard to a separate function and use it when unlit is true
## Notes
I tried calling `alpha_discard()` before the `if` in pbr.wgsl, but I had errors related to having a `discard` at the beginning before doing the texture sampling. I'm not sure if there's a way to fix that instead of having the function being called in 2 places.
# Objective
- Currently, errors aren't logged as soon as they are found, they are logged only on the next frame. This means your shader could have an unreported error that could have been reported on the first frame.
## Solution
- Log the error as soon as they are found, don't wait until next frame
## Notes
I discovered this issue because I was simply unwrapping the `Result` from `PipelinCache::get_render_pipeline()` which caused it to fail without any explanations. Admittedly, this was a bit of a user error, I shouldn't have unwrapped that, but it seems a bit strange to wait until the next time the pipeline is processed to log the error instead of just logging it as soon as possible since we already have all the info necessary.
# Objective
- Add ability to create nested spawns. This is needed for stageless. The current executor spawns tasks for each system early and runs the system by communicating through a channel. In stageless we want to spawn the task late, so that archetypes can be updated right before the task is run. The executor is run on a separate task, so this enables the scope to be passed to the spawned executor.
- Fixes#4301
## Solution
- Instantiate a single threaded executor on the scope and use that instead of the LocalExecutor. This allows the scope to be Send, but still able to spawn tasks onto the main thread the scope is run on. This works because while systems can access nonsend data. The systems themselves are Send. Because of this change we lose the ability to spawn nonsend tasks on the scope, but I don't think this is being used anywhere. Users would still be able to use spawn_local on TaskPools.
- Steals the lifetime tricks the `std:🧵:scope` uses to allow nested spawns, but disallow scope to be passed to tasks or threads not associated with the scope.
- Change the storage for the tasks to a `ConcurrentQueue`. This is to allow a &Scope to be passed for spawning instead of a &mut Scope. `ConcurrentQueue` was chosen because it was already in our dependency tree because `async_executor` depends on it.
- removed the optimizations for 0 and 1 spawned tasks. It did improve those cases, but made the cases of more than 1 task slower.
---
## Changelog
Add ability to nest spawns
```rust
fn main() {
let pool = TaskPool::new();
pool.scope(|scope| {
scope.spawn(async move {
// calling scope.spawn from an spawn task was not possible before
scope.spawn(async move {
// do something
});
});
})
}
```
## Migration Guide
If you were using explicit lifetimes and Passing Scope you'll need to specify two lifetimes now.
```rust
fn scoped_function<'scope>(scope: &mut Scope<'scope, ()>) {}
// should become
fn scoped_function<'scope>(scope: &Scope<'_, 'scope, ()>) {}
```
`scope.spawn_local` changed to `scope.spawn_on_scope` this should cover cases where you needed to run tasks on the local thread, but does not cover spawning Nonsend Futures.
## TODO
* [x] think real hard about all the lifetimes
* [x] add doc about what 'env and 'scope mean.
* [x] manually check that the single threaded task pool still works
* [x] Get updated perf numbers
* [x] check and make sure all the transmutes are necessary
* [x] move commented out test into a compile fail test
* [x] look through the tests for scope on std and see if I should add any more tests
Co-authored-by: Michael Hsu <myhsu@benjaminelectric.com>
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
Make `Res` cloneable
## Solution
Add an associated fn `clone(self: &Self) -. Self` instead of `Copy + Clone` trait impls to avoid `res.clone()` failing to clone out the underlying `T`
# Objective
Often one wants to create a `UiRect` with a value only specifying a single field. These ways are already available, but not the most ergonomic:
```rust
UiRect::new(Val::Undefined, Val::Undefined, Val::Percent(25.0), Val::Undefined)
```
```rust
UiRect {
top: Val::Percent(25.0),
..default()
}
```
## Solution
Introduce 6 new constructors:
- `horizontal`
- `vertical`
- `left`
- `right`
- `top`
- `bottom`
So the above code can be written instead as:
```rust
UiRect::top(Val::Percent(25.0))
```
This solution is similar to the style fields `margin-left`, `padding-top`, etc. that you would see in CSS, from which bevy's UI has other inspiration. Therefore, it should still feel intuitive to users coming from CSS.
---
## Changelog
### Added
- Additional constructors for `UiRect` to specify values for specific fields
# Objective
Currently, arrays cannot indexed using the reflection path API.
This change makes them behave like lists so `x.get_path("list[0]")` will behave the same way, whether x.list is a "List" (e.g. a Vec) or an array.
## Solution
When syntax is encounterd `[ <idx> ]` we check if the referenced type is either a `ReflectRef::List` or `ReflectRef::Array` (or `ReflectMut` for the mutable case). Since both provide the identical API for accessing entries, we do the same for both, although it requires code duplication as far as I can tell.
This was born from working on #5764, but since this seems to be an easier fix (and I am not sure if I can actually solve #5812) I figured it might be worth to split this out.
# Objective
Simple docs/comments only PR that just fixes some outdated file references left over from the render rewrite.
## Solution
- Change the references to point to the correct files
# Objective
The [Stageless RFC](https://github.com/bevyengine/rfcs/pull/45) involves allowing exclusive systems to be referenced and ordered relative to parallel systems. We've agreed that unifying systems under `System` is the right move.
This is an alternative to #4166 (see rationale in the comments I left there). Note that this builds on the learnings established there (and borrows some patterns).
## Solution
This unifies parallel and exclusive systems under the shared `System` trait, removing the old `ExclusiveSystem` trait / impls. This is accomplished by adding a new `ExclusiveFunctionSystem` impl similar to `FunctionSystem`. It is backed by `ExclusiveSystemParam`, which is similar to `SystemParam`. There is a new flattened out SystemContainer api (which cuts out a lot of trait and type complexity).
This means you can remove all cases of `exclusive_system()`:
```rust
// before
commands.add_system(some_system.exclusive_system());
// after
commands.add_system(some_system);
```
I've also implemented `ExclusiveSystemParam` for `&mut QueryState` and `&mut SystemState`, which makes this possible in exclusive systems:
```rust
fn some_exclusive_system(
world: &mut World,
transforms: &mut QueryState<&Transform>,
state: &mut SystemState<(Res<Time>, Query<&Player>)>,
) {
for transform in transforms.iter(world) {
println!("{transform:?}");
}
let (time, players) = state.get(world);
for player in players.iter() {
println!("{player:?}");
}
}
```
Note that "exclusive function systems" assume `&mut World` is present (and the first param). I think this is a fair assumption, given that the presence of `&mut World` is what defines the need for an exclusive system.
I added some targeted SystemParam `static` constraints, which removed the need for this:
``` rust
fn some_exclusive_system(state: &mut SystemState<(Res<'static, Time>, Query<&'static Player>)>) {}
```
## Related
- #2923
- #3001
- #3946
## Changelog
- `ExclusiveSystem` trait (and implementations) has been removed in favor of sharing the `System` trait.
- `ExclusiveFunctionSystem` and `ExclusiveSystemParam` were added, enabling flexible exclusive function systems
- `&mut SystemState` and `&mut QueryState` now implement `ExclusiveSystemParam`
- Exclusive and parallel System configuration is now done via a unified `SystemDescriptor`, `IntoSystemDescriptor`, and `SystemContainer` api.
## Migration Guide
Calling `.exclusive_system()` is no longer required (or supported) for converting exclusive system functions to exclusive systems:
```rust
// Old (0.8)
app.add_system(some_exclusive_system.exclusive_system());
// New (0.9)
app.add_system(some_exclusive_system);
```
Converting "normal" parallel systems to exclusive systems is done by calling the exclusive ordering apis:
```rust
// Old (0.8)
app.add_system(some_system.exclusive_system().at_end());
// New (0.9)
app.add_system(some_system.at_end());
```
Query state in exclusive systems can now be cached via ExclusiveSystemParams, which should be preferred for clarity and performance reasons:
```rust
// Old (0.8)
fn some_system(world: &mut World) {
let mut transforms = world.query::<&Transform>();
for transform in transforms.iter(world) {
}
}
// New (0.9)
fn some_system(world: &mut World, transforms: &mut QueryState<&Transform>) {
for transform in transforms.iter(world) {
}
}
```
# Objective
I was working with the TextBundle component bundle because I wanted to change the position of the text that the bundle was holding. I used the transform field on the TextBundle at first because that is normally what controls the position of sprites in Bevy and that's what I was used to working with.
But the actual way to change the position of text inside of a TextBundle is to use the Style's position field, not the TextBundle's transform field.
Anecdotally, it was mentioned on the discord that other users have had this issue too.
## Solution
I added a small doc comment to the TextBundle's transform telling users not to use it to set the position of text. And since this issue applies to the other UI bundles, I added comments there as well!
# Objective
Fixes#6078. The `UiColor` component is unhelpfully named: it is unclear, ambiguous with border color and
## Solution
Rename the `UiColor` component (and associated fields) to `BackgroundColor` / `background_colorl`.
## Migration Guide
`UiColor` has been renamed to `BackgroundColor`. This change affects `NodeBundle`, `ButtonBundle` and `ImageBundle`. In addition, the corresponding field on `ExtractedUiNode` has been renamed to `background_color` for consistency.
This is an adoption of #3775
This merges `TextureAtlas` `from_grid_with_padding` into `from_grid` , adding optional padding and optional offset.
Since the orignal PR, the offset had already been added to from_grid_with_padding through #4836
## Changelog
- Added `padding` and `offset` arguments to `TextureAtlas::from_grid`
- Removed `TextureAtlas::from_grid_with_padding`
## Migration Guide
`TextureAtlas::from_grid_with_padding` was merged into `from_grid` which takes two additional parameters for padding and an offset.
```
// 0.8
TextureAtlas::from_grid(texture_handle, Vec2::new(24.0, 24.0), 7, 1);
// 0.9
TextureAtlas::from_grid(texture_handle, Vec2::new(24.0, 24.0), 7, 1, None, None)
// 0.8
TextureAtlas::from_grid_with_padding(texture_handle, Vec2::new(24.0, 24.0), 7, 1, Vec2::new(4.0, 4.0));
// 0.9
TextureAtlas::from_grid(texture_handle, Vec2::new(24.0, 24.0), 7, 1, Some(Vec2::new(4.0, 4.0)), None)
```
Co-authored-by: olefish <88390729+oledfish@users.noreply.github.com>
# Objective
- Sometimes, like when using shaders, you can only use a time value in `f32`. Unfortunately this suffers from floating precision issues pretty quickly. The standard approach to this problem is to wrap the time after a given period
- This is necessary for https://github.com/bevyengine/bevy/pull/5409
## Solution
- Add a `seconds_since_last_wrapping_period` method on `Time` that returns a `f32` that is the `seconds_since_startup` modulo the `max_wrapping_period`
---
## Changelog
Added `seconds_since_last_wrapping_period` to `Time`
## Additional info
I'm very opened to hearing better names. I don't really like the current naming, I just went with something descriptive.
Co-authored-by: Charles <IceSentry@users.noreply.github.com>
# Objective
Now that we can consolidate Bundles and Components under a single insert (thanks to #2975 and #6039), almost 100% of world spawns now look like `world.spawn().insert((Some, Tuple, Here))`. Spawning an entity without any components is an extremely uncommon pattern, so it makes sense to give spawn the "first class" ergonomic api. This consolidated api should be made consistent across all spawn apis (such as World and Commands).
## Solution
All `spawn` apis (`World::spawn`, `Commands:;spawn`, `ChildBuilder::spawn`, and `WorldChildBuilder::spawn`) now accept a bundle as input:
```rust
// before:
commands
.spawn()
.insert((A, B, C));
world
.spawn()
.insert((A, B, C);
// after
commands.spawn((A, B, C));
world.spawn((A, B, C));
```
All existing instances of `spawn_bundle` have been deprecated in favor of the new `spawn` api. A new `spawn_empty` has been added, replacing the old `spawn` api.
By allowing `world.spawn(some_bundle)` to replace `world.spawn().insert(some_bundle)`, this opened the door to removing the initial entity allocation in the "empty" archetype / table done in `spawn()` (and subsequent move to the actual archetype in `.insert(some_bundle)`).
This improves spawn performance by over 10%:
To take this measurement, I added a new `world_spawn` benchmark.
Unfortunately, optimizing `Commands::spawn` is slightly less trivial, as Commands expose the Entity id of spawned entities prior to actually spawning. Doing the optimization would (naively) require assurances that the `spawn(some_bundle)` command is applied before all other commands involving the entity (which would not necessarily be true, if memory serves). Optimizing `Commands::spawn` this way does feel possible, but it will require careful thought (and maybe some additional checks), which deserves its own PR. For now, it has the same performance characteristics of the current `Commands::spawn_bundle` on main.
**Note that 99% of this PR is simple renames and refactors. The only code that needs careful scrutiny is the new `World::spawn()` impl, which is relatively straightforward, but it has some new unsafe code (which re-uses battle tested BundlerSpawner code path).**
---
## Changelog
- All `spawn` apis (`World::spawn`, `Commands:;spawn`, `ChildBuilder::spawn`, and `WorldChildBuilder::spawn`) now accept a bundle as input
- All instances of `spawn_bundle` have been deprecated in favor of the new `spawn` api
- World and Commands now have `spawn_empty()`, which is equivalent to the old `spawn()` behavior.
## Migration Guide
```rust
// Old (0.8):
commands
.spawn()
.insert_bundle((A, B, C));
// New (0.9)
commands.spawn((A, B, C));
// Old (0.8):
commands.spawn_bundle((A, B, C));
// New (0.9)
commands.spawn((A, B, C));
// Old (0.8):
let entity = commands.spawn().id();
// New (0.9)
let entity = commands.spawn_empty().id();
// Old (0.8)
let entity = world.spawn().id();
// New (0.9)
let entity = world.spawn_empty();
```
# Objective
- Add unit tests for ambiguity detection reporting.
- Incremental implementation of #4299.
## Solution
- Refactor ambiguity detection internals to make it testable. As a bonus, this should make it easier to extend in the future.
## Notes
* This code was copy-pasted from #4299 and modified. Credit goes to @alice-i-cecile and @afonsolage, though I'm not sure who wrote what at this point.
## Objective
Fixes https://github.com/bevyengine/bevy/issues/6063
## Solution
- Use `then_some(x)` instead of `then( || x)`.
- Updated error logs from `bevy_ecs_compile_fail_tests`.
## Migration Guide
From Rust 1.63 to 1.64, a new Clippy error was added; now one should use `then_some(x)` instead of `then( || x)`.
# Objective
Both components already derives `Reflect` and it would be nice to have `FromReflect` in order to ser/de between those types without relaying on `downcast`, since it can fail between different platforms, like WebAssembly.
## Solution
Derive `FromReflect` for `Transform` and `GlobalTransform`.
I thought if I should also derive `FromReflect` for `GlobalTransform`, since it's a computed component, but there may be some use cases where a `GlobalTransform` is needed to be sent over the wire, so I decided to do it.
# Objective
- Reconfigure surface after present mode changes. It seems that this is not done currently at runtime. It's pretty common for games to change such graphical settings at runtime.
- Fixes present mode issue in #5111
## Solution
- Exactly like resolution change gets tracked when extracting window, do the same for present mode.
Additionally, I added present mode (vsync) toggling to window settings example.
# Objective
Take advantage of the "impl Bundle for Component" changes in #2975 / add the follow up changes discussed there.
## Solution
- Change `insert` and `remove` to accept a Bundle instead of a Component (for both Commands and World)
- Deprecate `insert_bundle`, `remove_bundle`, and `remove_bundle_intersection`
- Add `remove_intersection`
---
## Changelog
- Change `insert` and `remove` now accept a Bundle instead of a Component (for both Commands and World)
- `insert_bundle` and `remove_bundle` are deprecated
## Migration Guide
Replace `insert_bundle` with `insert`:
```rust
// Old (0.8)
commands.spawn().insert_bundle(SomeBundle::default());
// New (0.9)
commands.spawn().insert(SomeBundle::default());
```
Replace `remove_bundle` with `remove`:
```rust
// Old (0.8)
commands.entity(some_entity).remove_bundle::<SomeBundle>();
// New (0.9)
commands.entity(some_entity).remove::<SomeBundle>();
```
Replace `remove_bundle_intersection` with `remove_intersection`:
```rust
// Old (0.8)
world.entity_mut(some_entity).remove_bundle_intersection::<SomeBundle>();
// New (0.9)
world.entity_mut(some_entity).remove_intersection::<SomeBundle>();
```
Consider consolidating as many operations as possible to improve ergonomics and cut down on archetype moves:
```rust
// Old (0.8)
commands.spawn()
.insert_bundle(SomeBundle::default())
.insert(SomeComponent);
// New (0.9) - Option 1
commands.spawn().insert((
SomeBundle::default(),
SomeComponent,
))
// New (0.9) - Option 2
commands.spawn_bundle((
SomeBundle::default(),
SomeComponent,
))
```
## Next Steps
Consider changing `spawn` to accept a bundle and deprecate `spawn_bundle`.
# Objective
The doc comments for `Command` methods are a bit inconsistent on the format, they sometimes go out of scope, and most importantly they are wrong, in the sense that they claim to perform the action described by the command, while in reality, they just push a command to perform the action.
- Follow-up of #5938.
- Related to #5913.
## Solution
- Where applicable, only stated that a `Command` is pushed.
- Added a “See also” section for similar methods.
- Added a missing “Panics” section for `Commands::entity`.
- Removed a wrong comment about `Commands::get_or_spawn` returning `None` (It does not return an option).
- Removed polluting descriptions of other items.
- Misc formatting changes.
## Future possibilities
Since the `Command` implementors (`Spawn`, `InsertBundle`, `InitResource`, ...) are public, I thought that it might be appropriate to describe the action of the command there instead of the method, and to add a `method → command struct` link to fill the gap.
If that seems too far-fetched, we may opt to make them private, if possible, or `#[doc(hidden)]`.
# Objective
Working on issue #1934 , with linking examples to the documentation. PR for transform examples.
## Solution
Added to the documentation in bevy_transform transform.rs and global_transform.rs utilizing links from examples.
[X] 3d_rotations.rs linked to rotate in Transform
[X] global_vs_local_translation.rs linked to top of Transform and GlobalTransform documentation
[X] scale.rs linked to scale Struct in Transform
[X] transform.rs linked to top of Transform documentation
[X] translation.rs linked to from_translation in Transform
Co-authored-by: bwhitt7 <103079612+bwhitt7@users.noreply.github.com>
@BoxyUwU this is your fault.
Also cart didn't arrive in time to tell us not to do this.
# Objective
- Fix#2974
## Solution
- The first commit just does the actual change
- Follow up commits do steps to prove that this method works to unify as required, but this does not remove `insert_bundle`.
## Changelog
### Changed
Nested bundles now collapse automatically, and every `Component` now implements `Bundle`.
This means that you can combine bundles and components arbitrarily, for example:
```rust
// before:
.insert(A).insert_bundle(MyBBundle{..})
// after:
.insert_bundle((A, MyBBundle {..}))
```
Note that there will be a follow up PR that removes the current `insert` impl and renames `insert_bundle` to `insert`.
### Removed
The `bundle` attribute in `derive(Bundle)`.
## Migration guide
In `derive(Bundle)`, the `bundle` attribute has been removed. Nested bundles are not collapsed automatically. You should remove `#[bundle]` attributes.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
> Note: This is rebased off #4561 and can be viewed as a competitor to that PR. See `Comparison with #4561` section for details.
# Objective
The current serialization format used by `bevy_reflect` is both verbose and error-prone. Taking the following structs[^1] for example:
```rust
// -- src/inventory.rs
#[derive(Reflect)]
struct Inventory {
id: String,
max_storage: usize,
items: Vec<Item>
}
#[derive(Reflect)]
struct Item {
name: String
}
```
Given an inventory of a single item, this would serialize to something like:
```rust
// -- assets/inventory.ron
{
"type": "my_game::inventory::Inventory",
"struct": {
"id": {
"type": "alloc::string::String",
"value": "inv001",
},
"max_storage": {
"type": "usize",
"value": 10
},
"items": {
"type": "alloc::vec::Vec<alloc::string::String>",
"list": [
{
"type": "my_game::inventory::Item",
"struct": {
"name": {
"type": "alloc::string::String",
"value": "Pickaxe"
},
},
},
],
},
},
}
```
Aside from being really long and difficult to read, it also has a few "gotchas" that users need to be aware of if they want to edit the file manually. A major one is the requirement that you use the proper keys for a given type. For structs, you need `"struct"`. For lists, `"list"`. For tuple structs, `"tuple_struct"`. And so on.
It also ***requires*** that the `"type"` entry come before the actual data. Despite being a map— which in programming is almost always orderless by default— the entries need to be in a particular order. Failure to follow the ordering convention results in a failure to deserialize the data.
This makes it very prone to errors and annoyances.
## Solution
Using #4042, we can remove a lot of the boilerplate and metadata needed by this older system. Since we now have static access to type information, we can simplify our serialized data to look like:
```rust
// -- assets/inventory.ron
{
"my_game::inventory::Inventory": (
id: "inv001",
max_storage: 10,
items: [
(
name: "Pickaxe"
),
],
),
}
```
This is much more digestible and a lot less error-prone (no more key requirements and no more extra type names).
Additionally, it is a lot more familiar to users as it follows conventional serde mechanics. For example, the struct is represented with `(...)` when serialized to RON.
#### Custom Serialization
Additionally, this PR adds the opt-in ability to specify a custom serde implementation to be used rather than the one created via reflection. For example[^1]:
```rust
// -- src/inventory.rs
#[derive(Reflect, Serialize)]
#[reflect(Serialize)]
struct Item {
#[serde(alias = "id")]
name: String
}
```
```rust
// -- assets/inventory.ron
{
"my_game::inventory::Inventory": (
id: "inv001",
max_storage: 10,
items: [
(
id: "Pickaxe"
),
],
),
},
```
By allowing users to define their own serialization methods, we do two things:
1. We give more control over how data is serialized/deserialized to the end user
2. We avoid having to re-define serde's attributes and forcing users to apply both (e.g. we don't need a `#[reflect(alias)]` attribute).
### Improved Formats
One of the improvements this PR provides is the ability to represent data in ways that are more conventional and/or familiar to users. Many users are familiar with RON so here are some of the ways we can now represent data in RON:
###### Structs
```js
{
"my_crate::Foo": (
bar: 123
)
}
// OR
{
"my_crate::Foo": Foo(
bar: 123
)
}
```
<details>
<summary>Old Format</summary>
```js
{
"type": "my_crate::Foo",
"struct": {
"bar": {
"type": "usize",
"value": 123
}
}
}
```
</details>
###### Tuples
```js
{
"(f32, f32)": (1.0, 2.0)
}
```
<details>
<summary>Old Format</summary>
```js
{
"type": "(f32, f32)",
"tuple": [
{
"type": "f32",
"value": 1.0
},
{
"type": "f32",
"value": 2.0
}
]
}
```
</details>
###### Tuple Structs
```js
{
"my_crate::Bar": ("Hello World!")
}
// OR
{
"my_crate::Bar": Bar("Hello World!")
}
```
<details>
<summary>Old Format</summary>
```js
{
"type": "my_crate::Bar",
"tuple_struct": [
{
"type": "alloc::string::String",
"value": "Hello World!"
}
]
}
```
</details>
###### Arrays
It may be a bit surprising to some, but arrays now also use the tuple format. This is because they essentially _are_ tuples (a sequence of values with a fixed size), but only allow for homogenous types. Additionally, this is how RON handles them and is probably a result of the 32-capacity limit imposed on them (both by [serde](https://docs.rs/serde/latest/serde/trait.Serialize.html#impl-Serialize-for-%5BT%3B%2032%5D) and by [bevy_reflect](https://docs.rs/bevy/latest/bevy/reflect/trait.GetTypeRegistration.html#impl-GetTypeRegistration-for-%5BT%3B%2032%5D)).
```js
{
"[i32; 3]": (1, 2, 3)
}
```
<details>
<summary>Old Format</summary>
```js
{
"type": "[i32; 3]",
"array": [
{
"type": "i32",
"value": 1
},
{
"type": "i32",
"value": 2
},
{
"type": "i32",
"value": 3
}
]
}
```
</details>
###### Enums
To make things simple, I'll just put a struct variant here, but the style applies to all variant types:
```js
{
"my_crate::ItemType": Consumable(
name: "Healing potion"
)
}
```
<details>
<summary>Old Format</summary>
```js
{
"type": "my_crate::ItemType",
"enum": {
"variant": "Consumable",
"struct": {
"name": {
"type": "alloc::string::String",
"value": "Healing potion"
}
}
}
}
```
</details>
### Comparison with #4561
This PR is a rebased version of #4561. The reason for the split between the two is because this PR creates a _very_ different scene format. You may notice that the PR descriptions for either PR are pretty similar. This was done to better convey the changes depending on which (if any) gets merged first. If #4561 makes it in first, I will update this PR description accordingly.
---
## Changelog
* Re-worked serialization/deserialization for reflected types
* Added `TypedReflectDeserializer` for deserializing data with known `TypeInfo`
* Renamed `ReflectDeserializer` to `UntypedReflectDeserializer`
* ~~Replaced usages of `deserialize_any` with `deserialize_map` for non-self-describing formats~~ Reverted this change since there are still some issues that need to be sorted out (in a separate PR). By reverting this, crates like `bincode` can throw an error when attempting to deserialize non-self-describing formats (`bincode` results in `DeserializeAnyNotSupported`)
* Structs, tuples, tuple structs, arrays, and enums are now all de/serialized using conventional serde methods
## Migration Guide
* This PR reduces the verbosity of the scene format. Scenes will need to be updated accordingly:
```js
// Old format
{
"type": "my_game::item::Item",
"struct": {
"id": {
"type": "alloc::string::String",
"value": "bevycraft:stone",
},
"tags": {
"type": "alloc::vec::Vec<alloc::string::String>",
"list": [
{
"type": "alloc::string::String",
"value": "material"
},
],
},
}
// New format
{
"my_game::item::Item": (
id: "bevycraft:stone",
tags: ["material"]
)
}
```
[^1]: Some derives omitted for brevity.
# Objective
Add traits to events in `bevy_input` and `bevy_windows`: `Copy`, `Serialize`/`Deserialize`, `PartialEq`, and `Eq`, as requested in https://github.com/bevyengine/bevy/issues/6022, https://github.com/bevyengine/bevy/issues/6023, https://github.com/bevyengine/bevy/issues/6024.
## Solution
Added the traits to events in `bevy_input` and `bevy_windows`. Added dependency of `serde` in `Cargo.toml` of `bevy_input`.
## Migration Guide
If one has been `.clone()`'ing `bevy_input` events, Clippy will now complain about that. Just remove `.clone()` to solve.
## Other Notes
Some events in `bevy_input` had `f32` fields, so `Eq` trait was not derived for them.
Some events in `bevy_windows` had `String` fields, so `Copy` trait was not derived for them.
Co-authored-by: targrub <62773321+targrub@users.noreply.github.com>
# Objective
Implement `IntoIterator` for `&Extract<P>` if the system parameter it wraps implements `IntoIterator`.
Enables the use of `IntoIterator` with an extracted query.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
`AssetServer::watch_for_changes()` is racy and redundant with `AssetServerSettings`.
Closes#5964.
## Changelog
* Remove `AssetServer::watch_for_changes()`
* Add `AssetServerSettings` to the prelude.
* Minor cleanup.
## Migration Guide
`AssetServer::watch_for_changes()` was removed.
Instead, use the `AssetServerSettings` resource.
```rust
app // AssetServerSettings must be inserted before adding the AssetPlugin or DefaultPlugins.
.insert_resource(AssetServerSettings {
watch_for_changes: true,
..default()
})
```
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
When trying derive `Debug` for type that has `DynamicEnum` it wasn't possible, since neither of `DynamicEnum`, `DynamicTuple`, `DynamicVariant` or `DynamicArray` implements `Debug`.
## Solution
Implement Debug for those types, using `derive` macro
---
## Changelog
- `DynamicEnum`, `DynamicTuple`, `DynamicVariant` and `DynamicArray` now implements `Debug`
# Objective
Fixes#5636
Summary: The FontAtlasSet caches generated font textures per font size. Since font size can be any arbitrary floating point number it is possible for the user to generate thousands of font texture inadvertently by changing the font size over time. This results in a memory leak as these generated font textures fill the available memory.
## Solution
We limit the number of possible font sizes that we will cache and throw an error if the user attempts to generate more. This error encourages the user to use alternative, less performance intensive methods to accomplish the same goal. If the user requires more font sizes and the alternative solutions wont work there is now a TextSettings Resource that the user can set to configure this limit.
---
## Changelog
The number of cached font sizes per font is now limited with a default limit of 100 font sizes per font. This limit is configurable via the new TextSettings struct.
# Objective
A common pitfall since 0.8 is the requirement on `ComputedVisibility`
being present on all ancestors of an entity that itself has
`ComputedVisibility`, without which, the entity becomes invisible.
I myself hit the issue and got very confused, and saw a few people hit
it as well, so it makes sense to provide a hint of what to do when such
a situation is encountered.
- Fixes#5849
- Closes#5616
- Closes#2277
- Closes#5081
## Solution
We now check that all entities with both a `Parent` and a
`ComputedVisibility` component have parents that themselves have a
`ComputedVisibility` component.
Note that the warning is only printed once.
We also add a similar warning to `GlobalTransform`.
This only emits a warning. Because sometimes it could be an intended
behavior.
Alternatives:
- Do nothing and keep repeating to newcomers how to avoid recurring
pitfalls
- Make the transform and visibility propagation tolerant to missing
components (#5616)
- Probably archetype invariants, though the current draft would not
allow detecting that kind of errors
---
## Changelog
- Add a warning when encountering dubious component hierarchy structure
Co-authored-by: Nicola Papale <nicopap@users.noreply.github.com>
# Objective
- To address problems outlined in https://github.com/bevyengine/bevy/issues/5245
## Solution
- Introduce `reflect(skip_serializing)` on top of `reflect(ignore)` which disables automatic serialisation to scenes, but does not disable reflection of the field.
---
## Changelog
- Adds:
- `bevy_reflect::serde::type_data` module
- `SerializationData` structure for describing which fields are to be/not to be ignored, automatically registers as type_data for struct-based types
- the `skip_serialization` flag for `#[reflect(...)]`
- Removes:
- ability to ignore Enum variants in serialization, since that didn't work anyway
## Migration Guide
- Change `#[reflect(ignore)]` to `#[reflect(skip_serializing)]` where disabling reflection is not the intended effect.
- Remove ignore/skip attributes from enum variants as these won't do anything anymore
# Objective
As of Rust 1.59, `std:🧵:available_parallelism` has been stabilized. As of Rust 1.61, the API matches `num_cpus::get` by properly handling Linux's cgroups and other sandboxing mechanisms.
As bevy does not have an established MSRV, we can replace `num_cpus` in `bevy_tasks` and reduce our dependency tree by one dep.
## Solution
Replace `num_cpus` with `std:🧵:available_parallelism`. Wrap it to have a fallback in the case it errors out and have it operate in the same manner as `num_cpus` did.
This however removes `physical_core_count` from the API, though we are currently not using it in any way in first-party crates.
---
## Changelog
Changed: `bevy_tasks::logical_core_count` -> `bevy_tasks::available_parallelism`.
Removed: `bevy_tasks::physical_core_count`.
## Migration Guide
`bevy_tasks::logical_core_count` and `bevy_tasks::physical_core_count` have been removed. `logical_core_count` has been replaced with `bevy_tasks::available_parallelism`, which works identically. If `bevy_tasks::physical_core_count` is required, the `num_cpus` crate can be used directly, as these two were just aliases for `num_cpus` APIs.
# Objective
Fixes Issue #6005.
## Solution
Replaced WorldQuery with ReadOnlyWorldQuery on F generic in Query filters and QueryState to restrict its trait bound.
## Migration Guide
Query filter (`F`) generics are now bound by `ReadOnlyWorldQuery`, rather than `WorldQuery`. If for some reason you were requesting `Query<&A, &mut B>`, please use `Query<&A, With<B>>` instead.
Very small change that improves the usability of `Sprite`.
Before this PR, the only way to render a portion of an `Image` was to create a `TextureAtlas` and use `TextureAtlasSprite`/`SpriteSheetBundle`. This can be very annoying for one-off use cases, like if you just want to remove a border from an image, or something. Using `Sprite`/`SpriteBundle` always meant that the entire full image would be rendered.
This PR adds an optional `rect` field to `Sprite`, allowing a sub-rectangle of the image to be rendered. This is similar to how texture atlases work, but does not require creating a texture atlas asset, making it much more convenient and efficient for quick one-off use cases.
Given how trivial this change is, it really felt like missing functionality in Bevy's sprites API. ;)
## Changelog
Added:
- `rect` field on `Sprite`: allows rendering a portion of the sprite's image; more convenient for one-off use cases, than creating a texture atlas.
# Objective
fixes#5946
## Solution
adjust cluster index calculation for viewport origin.
from reading point 2 of the rasterization algorithm description in https://gpuweb.github.io/gpuweb/#rasterization, it looks like framebuffer space (and so @bulitin(position)) is not meant to be adjusted for viewport origin, so we need to subtract that to get the right cluster index.
- add viewport origin to rust `ExtractedView` and wgsl `View` structs
- subtract from frag coord for cluster index calculation
# Objective
Currently some TextureFormats are not supported by the Image type.
The `TextureFormat::Rg16Unorm` format is useful for storing minmax heightmaps.
Similar to #5249 I now additionally require image to support the dual channel variant.
## Solution
Added `TextureFormat::Rg16Unorm` support to Image.
Additionally this PR derives `Resource` for `SpecializedComputePipelines`, because for some reason this was missing.
All other special pipelines do derive `Resource` already.
Co-authored-by: Kurt Kühnert <51823519+Ku95@users.noreply.github.com>
# Objective
Fixes#5963
## Solution
Add remaining fn in Timer class, this function only minus total duration with elapsed time.
Co-authored-by: Sergi-Ferrez <61662926+Sergi-Ferrez@users.noreply.github.com>
# Objective
While coding in bevy I needed to get the elapsed time of a stopwatch as f64.
I found it quite odd there are functions on Timer to get time as f64 but not on the Stopwatch.
## Solution
- added a function that returns the `Stopwatch` elapsed time as `f64`
---
## Changelog
### Added
- Added a function to get `Stopwatch` elapsed time as `f64`
### Fixed
- The Stopwatch elapsed function had a wrong docs link
# Objective
While using the ParallelExecutor, systems do not actually start until `prepare_systems` completes. In stages where there are large numbers of "empty" systems with very little work to do, this delay adds significant overhead, which can add up over many stages.
## Solution
Immediately and synchronously signal the start of systems that can run without dependencies inside `prepare_systems` instead of waiting for the first executor iteration after `prepare_systems` completes. Any system that is dependent on them still cannot run until after `prepare_systems` completes, but there are a large number of unconstrained systems in the base engine where this is a general benefit in almost every case.
## Performance
This change was tested against `many_foxes` in the default configuration. As this change is sensitive to the overhead around scheduling systems, the spans for measuring system timing, system overhead, and system commands were all commented out for these measurements.
The median stage timings between `main` and this PR are as follows:
|stage|main|this PR|
|:--|:--|:--|
|First|75.54 us|61.61 us|
|LoadAssets|51.05 us|42.32 us|
|PreUpdate|54.6 us|55.56 us|
|Update|61.89 us|51.5 us|
|PostUpdate|7.27 ms|6.71 ms|
|AssetEvents|47.82 us|35.95 us|
|Last|39.19 us|37.71 us|
|reserve_and_flush|57.83 us|48.2 us|
|Extract|1.41 ms|1.28 ms|
|Prepare|554.49 us|502.53 us|
|Queue|216.29 us|207.51 us|
|Sort|67.03 us|60.99 us|
|Render|1.73 ms|1.58 ms|
|Cleanup|33.55 us|30.76 us|
|Clear Entities|18.56 us|17.05 us|
|**full frame**|**11.9 ms**|**10.91 ms**|
For the first few stages, the benefit is small but cumulative over each. For PostUpdate in particular, this allows `parent_update` to run while prepare_systems is running, which is required for the animation and transform propagation systems, which dominate the time spent in the stage, but also frontloads the contention as the other "empty" systems are also running while `parent_update` is running. For Render, where there is just a single large exclusive system, the benefit comes from not waiting on a spuriously scheduled task on the task pool to kick off the system: it's immediately scheduled to run.
# Objective
EntityMut::world takes &mut self instead of &self I don't see any reason for this.
EntityRef is overly restrictive with fn world and could return &'w World
---
## Changelog
- EntityRef now implements Copy and Clone
- EntityRef::world is now fn(&self) -> &'w World instead of fn(&mut self) -> &World
- EntityMut::world is now fn(&self) -> &World instead of fn(&mut self) -> &World
# Objective
Currently, `Local` has a `Sync` bound. Theoretically this is unnecessary as a local can only ever be accessed from its own system, ensuring exclusive access on one thread. This PR removes this restriction.
## Solution
- By removing the `Resource` bound from `Local` and adding the new `SyncCell` threading primative, `Local` can have the `Sync` bound removed.
## Changelog
### Added
- Added `SyncCell` to `bevy_utils`
### Changed
- Removed `Resource` bound from `Local`
- `Local` is now wrapped in a `SyncCell`
## Migration Guide
- Any code relying on `Local<T>` having `T: Resource` may have to be changed, but this is unlikely.
Co-authored-by: PROMETHIA-27 <42193387+PROMETHIA-27@users.noreply.github.com>
While looking into `collide()`, I wrote some tests to confirm the behavior I read in the code. This PR adds those tests and improves the documentation.
Co-authored-by: robem <669201+robem@users.noreply.github.com>
# Objective
- Make people stop believing that commands are applied immediately (hopefully).
- Close#5913.
- Alternative to #5930.
## Solution
I added the clause “to perform impactful changes to the `World`” to the first line to subliminally help the reader accept the fact that some operations cannot be performed immediately without messing up everything.
Then I explicitely said that applying a command requires exclusive `World` access, and finally I proceeded to show when these commands are automatically applied.
I also added a brief paragraph about how commands can be applied manually, if they want.
---
### Further possibilities
If you agree, we can also change the text of the method documentation (in a separate PR) to stress about enqueueing an action instead of just performing it. For example, in `Commands::spawn`:
> Creates a new `Entity`
would be changed to something like:
> Issues a `Command` to spawn a new `Entity`
This may even have a greater effect, since when typing in an IDE, the docs of the method pop up and the programmer can read them on the fly.
# Objective
Without this we can inappropriately merge batches together without properly accounting for non-batch items between them, and the merged batch will then be sorted incorrectly later.
This change seems to reliably fix the issue I was seeing in #5919.
## Solution
Ensure the `batch_phase_system` runs after the `sort_phase_system`, so that batching can only look at actually adjacent phase items.
# Objective
I wanted to run the code
```rust
let reflect_resource: ReflectResource = ...;
let value: Mut<dyn Reflect> = reflect_resource.reflect(world);
value.deref();
// ^ ERROR: deref method doesn't exist because `dyn Reflect` doesnt satisfy `: Sized`.
```
## Solution
Relax `Sized` bounds in all the methods and trait implementations for `Mut` and friends.
# Objective
This code is very disjoint, and the `stage.rs` file that it's in is already very long.
All I've done is move the code and clean up the compiler errors that result.
Followup to #5916, split out from #4299.
# Objective
Ambiguity sets are used to ignore system order ambiguities between groups of systems. However, they are not very useful: they are clunky, poorly integrated, and generally hampered by the difficulty using (or discovering) the ambiguity detector.
As a first step to the work in #4299, we're removing them.
## Migration Guide
Ambiguity sets have been removed.
# Objective
- Our existing change detection API is not flexible enough for advanced users: particularly those attempting to do rollback networking.
- This is an important use case, and with adequate warnings we can make mucking about with change ticks scary enough that users generally won't do it.
- Fixes#5633.
- Closes#2363.
## Changelog
- added `ChangeDetection::set_last_changed` to manually mutate the `last_change_ticks` field"
- the `ChangeDetection` trait now requires an `Inner` associated type, which contains the value being wrapped.
- added `ChangeDetection::bypass_change_detection`, which hands out a raw `&mut Inner`
## Migration Guide
Add the `Inner` associated type and new methods to any type that you've implemented `DetectChanges` for.
# Objective
- I'm currently working on being able to call methods on reflect types (https://github.com/jakobhellermann/bevy_reflect_fns)
- for that, I'd like to add methods to the `Input<KeyCode>` resource (which I'm doing by registering type data)
- implementing `Reflect` is currently a requirement for having type data in the `TypeRegistry`
## Solution
- derive `Reflect` for `KeyCode` and `Input`
- uses `#[reflect_value]` for `Input`, since it's fields aren't supposed to be observable
- using reflect_value would need `Clone` bounds on `T`, but since all the methods (`.pressed` etc) already require `T: Copy`, I unified everything to requiring `Copy`
- add `Send + Sync + 'static` bounds, also required by reflect derive
## Unrelated improvements
I can extract into a separate PR if needed.
- the `Reflect` derive would previously ignore `#[reflect_value]` and only accept `#[reflect_value()]` which was a bit confusing
- the generated code used `val.clone()` on a reference, which is fine if `val` impls `Clone`, but otherwise also compiles with a worse error message. Change to `std::clone::Clone::clone(val)` instead which gives a neat `T does not implement Clone` error
Make API users aware that the type aliases `QueryItem` and `QueryFetch` can be used instead of the more bloated alternative with `WorldQueryGats`.
Fixes#5842
# Objective
Make `TextLayoutInfo` more accessible as a component, rather than internal to `TextPipeline`. I am working on a plugin that manipulates these and there is no (mutable) access to them right now.
## Solution
This changes `TextPipeline::queue_text` to return `TextLayoutInfo`'s rather than storing them in a map internally. `text2d_system` and `text_system` now take the returned `TextLayoutInfo` and store it as a component of the entity. I considered adding an accessor to `TextPipeline` (e.g. `get_glyphs_mut`) but this seems like it might be a little faster, and also has the added benefit of cleaning itself up when entities are removed. Right now nothing is ever removed from the glyphs map.
## Changelog
Removed `DefaultTextPipeline`. `TextPipeline` no longer has a generic key type. `TextPipeline::queue_text` returns `TextLayoutInfo` directly.
## Migration Guide
This might break a third-party crate? I could restore the orginal TextPipeline API as a wrapper around what's in this PR.
# Objective
Fixes#5882
## Solution
Per https://github.com/rust-windowing/winit/issues/1705, the root cause is "UIWindow should be created inside UIApplicationMain". Currently, there are two places to create UIWindow, one is Plugin's build function, which is not inside UIApplicationMain. Just comment it out, and it works.
# Objective
Support monitor selection for all window modes.
Fixes#5875.
## Changelog
* Moved `MonitorSelection` out of `WindowPosition::Centered`, into `WindowDescriptor`.
* `WindowPosition::At` is now relative to the monitor instead of being in 'desktop space'.
* Renamed `MonitorSelection::Number` to `MonitorSelection::Index` for clarity.
* Added `WindowMode` to the prelude.
* `Window::set_position` is now relative to a monitor and takes a `MonitorSelection` as argument.
## Migration Guide
`MonitorSelection` was moved out of `WindowPosition::Centered`, into `WindowDescriptor`.
`MonitorSelection::Number` was renamed to `MonitorSelection::Index`.
```rust
// Before
.insert_resource(WindowDescriptor {
position: WindowPosition::Centered(MonitorSelection::Number(1)),
..default()
})
// After
.insert_resource(WindowDescriptor {
monitor: MonitorSelection::Index(1),
position: WindowPosition::Centered,
..default()
})
```
`Window::set_position` now takes a `MonitorSelection` as argument.
```rust
window.set_position(MonitorSelection::Current, position);
```
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
Clean up taffy nodes when the associated UI node gets removed. The current UI code will keep the taffy nodes around forever.
## Solution
Use `RemovedComponents<Node>` to iterate over nodes that are no longer valid UI nodes or that have been despawned, and remove them from taffy and the internal hash map.
## Implementation Notes
Do note that using `despawn()` instead of `despawn_recursive()` on a UI node that has children will result in a [warnings spam](https://github.com/bevyengine/bevy/blob/main/crates/bevy_ui/src/flex/mod.rs#L120) since the children will not be part of a proper UI hierarchy anymore.
---
## Changelog
- Fixed memory leak when nodes are removed in bevy_ui
# Objective
- The `Gamepad` type is a tiny value-containing type that implements `Copy`.
- By convention, references to `Copy` types should be avoided, as they can introduce overhead and muddle the semantics of what's going on.
- This allows us to reduce boilerplate reference manipulation and lifetimes in user facing code.
## Solution
- Make assorted methods on `Gamepads` take / return a raw `Gamepad`, rather than `&Gamepad`.
## Migration Guide
- `Gamepads::iter` now returns an iterator of `Gamepad`. rather than an iterator of `&Gamepad`.
- `Gamepads::contains` now accepts a `Gamepad`, rather than a `&Gamepad`.
## Solution
Exposes the image <-> "texture" as methods on `Image`.
## Extra
I'm wondering if `image_texture_conversion.rs` should be renamed to `image_conversion.rs`. That or the file be deleted altogether in favour of putting the code alongside the rest of the `Image` impl. Its kind-of weird to refer to the `Image` as a texture.
Also `Image::convert` is a public method so I didn't want to edit its signature, but it might be nice to have the function consume the image instead of just passing a reference to it because it would eliminate a clone.
## Changelog
> Rename `image_to_texture` to `Image::from_dynamic`
> Rename `texture_to_image` to `Image::try_into_dynamic`
> `Image::try_into_dynamic` now returns a `Result` (this is to make it easier for users who didn't read that only a few conversions are supported to figure it out.)
# Objective
Document most of the public items of the `bevy_render::camera` module and its
sub-modules.
## Solution
Add docs to most public items. Follow-up from #3447.
# Objective
- Increase consistency across documentation of `Query` methods.
- Fixes#5506
## Solution
- See #4989. This PR is derived from it. It just includes changes to the `Query` methods' docs.
# Objective
Document `PipelineCache` and a few other related types.
## Solution
Add documenting comments to `PipelineCache` and a few other related
types in the same file.
# Objective
https://github.com/bevyengine/bevy/pull/503 added these.
I don't know what problem it solved, the PR doesn't say and the code didn't make it obvious to me.
## Solution
AFAIK removing unsafe `Send`/`Sync` impls can't introduce unsoundness.
Yeet.
## Migration Guide
Why tho.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
The documentation on `Reflect` doesn't account for the recently added reflection traits: [`Array`](https://github.com/bevyengine/bevy/pull/4701) and [`Enum`](https://github.com/bevyengine/bevy/pull/4761).
## Solution
Updated the documentation for `Reflect` to account for the `Array` and `Enum`.
Co-authored-by: Gino Valente <49806985+MrGVSV@users.noreply.github.com>
# Objective
- Update notify dependency to 5.0.0 stable
- Fix breaking changes
- Closes#5861
## Solution
- RecommendedWatcher now takes a Config argument. Giving it the default Config should be the same behavior as before (check every 30 seconds)
# Objective
- Update ron to 0.8.0
- Fix breaking changes
- Closes#5862
## Solution
- Removed now non-existing method call (behavior is now the same without it)
# Objective
- Update `Query` docs with better terminology
- add some performance remarks (Fixes#4742)
## Solution
- See #4989. This PR is derived from it. It just includes changes to the `Query` struct docs.
# Objective
Promote the `Rect` utility of `sprite::Rect`, which defines a rectangle
by its minimum and maximum corners, to the `bevy_math` crate to make it
available as a general math type to all crates without the need to
depend on the `bevy_sprite` crate.
Fixes#5575
## Solution
Move `sprite::Rect` into `bevy_math` and fix all uses.
Implement `Reflect` for `Rect` directly into the `bevy_reflect` crate by
having `bevy_reflect` depend on `bevy_math`. This looks like a new
dependency, but the `bevy_reflect` was "cheating" for other math types
by directly depending on `glam` to reflect other math types, thereby
giving the illusion that there was no dependency on `bevy_math`. In
practice conceptually Bevy's math types are reflected into the
`bevy_reflect` crate to avoid a dependency of that crate to a "lower
level" utility crate like `bevy_math` (which in turn would make
`bevy_reflect` be a dependency of most other crates, and increase the
risk of circular dependencies). So this change simply formalizes that
dependency in `Cargo.toml`.
The `Rect` struct is also augmented in this change with a collection of
utility methods to improve its usability. A few uses cases are updated
to use those new methods, resulting is more clear and concise syntax.
---
## Changelog
### Changed
- Moved the `sprite::Rect` type into `bevy_math`.
### Added
- Added several utility methods to the `math::Rect` type.
## Migration Guide
The `bevy::sprite::Rect` type moved to the math utility crate as
`bevy::math::Rect`. You should change your imports from `use
bevy::sprite::Rect` to `use bevy::math::Rect`.
# Objective
- In WASM, creating a pipeline can easily take 2 seconds, freezing the game while doing so
- Preloading pipelines can be done during a "loading" state, but it is not trivial to know which pipeline to preload, or when it's done
## Solution
- Add a log with shaders being loaded and their shader defs
- add a function on `PipelineCache` to return the number of ready pipelines
# Objective
- Easier to work with model assets
- Models are often one mesh, many textures. This can be hard to use in Bevy as it's not possible to clone the scene to have one scene for each material. It's still possible to instantiate the texture-less scene, then modify the texture material once spawned but that means happening during play and is quite more painful
## Solution
- Expose the code to clone a scene. This code already existed but was only possible to use to spawn the scene
# Objective
Extend the scope of Gamepad to accommodate devices that have more inputs than a typical controller.
## Solution
Add additional enum variants to both _GamepadButtonType_ and _GamepadAxisType_ that supports up to 255 more non-standard buttons/axis respectively.
## Personal motivation
I have been writing an alternative to the GILRS crate, and with this simple change to the source code, It will be a trivial thing to direct new devices through the bevy systems, even when they do not always behave exactly like your typical controller.
# Objective
- Fixes#5850
## Solution
- As described in the issue, added a `get_entity` method on `Commands` that returns an `Option<EntityCommands>`
## Changelog
- Added the new method with a simple doc test
- I have re-used `get_entity` in `entity`, similarly to how `get_single` is used in `single` while additionally preserving the error message
- Add `#[inline]` to both functions
Entities that have commands queued to despawn system will still return commands when `get_entity` is called but that is representative of the fact that the entity is still around until those commands are flushed.
A potential `contains_entity` could also be added in this PR if desired, that would effectively be replacing Entities.contains but may be more discoverable if this is a common use case.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
I'm build a UI system for bevy. In this UI system there is a concept of a system per UI entity. I had an issue where change detection wasn't working how I would expect and it's because when a function system is ran the `last_change_tick` is updated with the latest tick(from world). In my particular case I want to "wait" to update the `last_change_tick` until after my system runs for each entity.
## Solution
Initially I thought bypassing the change detection all together would be a good fix, but on talking to some users in discord a simpler fix is to just expose `last_change_tick` to the end users. This is achieved by adding the following to the `System` trait:
```rust
/// Allows users to get the system's last change tick.
fn get_last_change_tick(&self) -> u32;
/// Allows users to set the system's last change tick.
fn set_last_change_tick(&mut self, last_change_tick: u32);
```
This causes a bit of weirdness with two implementors of `System`. `FixedTimestep` and `ChainSystem` both implement system and thus it's required that some sort of implementation be given for the new functions. I solved this by outputting a warning and not doing anything for these systems.
I think it's important to understand why I can't add the new functions only to the function system and not to the `System` trait. In my code I store the systems generically as `Box<dyn System<...>>`. I do this because I have differing parameters that are being passed in depending on the UI widget's system. As far as I can tell there isn't a way to take a system trait and cast it into a specific type without knowing what those parameters are.
In my own code this ends up looking something like:
```rust
// Runs per entity.
let old_tick = widget_system.get_last_change_tick();
should_update_children = widget_system.run((widget_tree.clone(), entity.0), world);
widget_system.set_last_change_tick(old_tick);
// later on after all the entities have been processed:
for system in context.systems.values_mut() {
system.set_last_change_tick(world.read_change_tick());
}
```
## Changelog
- Added `get_last_change_tick` and `set_last_change_tick` to `System`'s.
# Objective
Since `identity` is a const fn that takes no arguments it seems logical to make it an associated constant.
This is also more in line with types from glam (eg. `Quat::IDENTITY`).
## Migration Guide
The method `identity()` on `Transform`, `GlobalTransform` and `TransformBundle` has been deprecated.
Use the associated constant `IDENTITY` instead.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
- `for_each` methods inconsistently used an actual generic param or `impl Trait` change it to use `impl Trait` always, change them to be consistent
- some methods returned `'w 's` or `'_ '_`, change them to return `'_ 's`
## Solution
- Do what i just said
---
## Changelog
- `iter_unsafe` and `get_unchecked` no longer return borrows tied to `'w`
## Migration Guide
transmute the returned borrow from `iter_unsafe` and `get_unchecked` if this broke you (although preferably find a way to write your code that doesnt need to do this...)
# Objective
Sometimes it's useful to be able to retrieve all the fields of a container type so that they may be processed separately. With reflection, however, we typically only have access to references.
The only alternative is to "clone" the value using `Reflect::clone_value`. This, however, returns a Dynamic type in most cases. The solution there would be to use `FromReflect` instead, but this also has a problem in that it means we need to add `FromReflect` as an additional bound.
## Solution
Add a `drain` method to all container traits. This returns a `Vec<Box<dyn Reflect>>` (except for `Map` which returns `Vec<(Box<dyn Reflect>, Box<dyn Reflect>)>`).
This allows us to do things a lot simpler. For example, if we finished processing a struct and just need a particular value:
```rust
// === OLD === //
/// May or may not return a Dynamic*** value (even if `container` wasn't a `DynamicStruct`)
fn get_output(container: Box<dyn Struct>, output_index: usize) -> Box<dyn Reflect> {
container.field_at(output_index).unwrap().clone_value()
}
// === NEW === //
/// Returns _exactly_ whatever was in the given struct
fn get_output(container: Box<dyn Struct>, output_index: usize) -> Box<dyn Reflect> {
container.drain().remove(output_index).unwrap()
}
```
### Discussion
* Is `drain` the best method name? It makes sense that it "drains" all the fields and that it consumes the container in the process, but I'm open to alternatives.
---
## Changelog
* Added a `drain` method to the following traits:
* `Struct`
* `TupleStruct`
* `Tuple`
* `Array`
* `List`
* `Map`
* `Enum`
# Objective
- The reflection `List` trait does not have a `pop` function.
- Popping elements off a list is a common use case and is almost always supported by `List`-like types.
## Solution
- Add the `pop()` method to the `List` trait and add the appropriate implementations of this function.
## Migration Guide
- Any custom type that implements the `List` trait will now need to implement the `pop` method.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
Fixes#5763
## Solution
Implemented as reflect value like the current `Range`. Is there a benefit to changing everything to a reflect struct?
# Objective
remove `insert_resource_with_id` because `insert_resource_by_id` exists and does almost exactly the same thing
blocked on #5587 because otherwise we will leak a resource when it's inserted
## Solution
remove the function and also add a safety invariant of to `insert_resource_by_id` that the id be valid for the world.
I didn't see any discussion in #4447 about this safety invariant being left off in favor of a panic so I'm curious if there was one or if it just seemed nicer to have less safety invariants for callers to uphold 😅
---
## Changelog
- safety invariant added to `insert_resource_by_id` requiring the id to be valid for world
## Migration Guide
- audit any calls to `insert_resource_by_id` making sure that the id is valid for the world
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
Fixes#5581
## Solution
`Window::scale_factor` already has logic for selecting the overridden or actual scale factor, so use it.
I tested this with the displays I have access to, but more testing would be nice. This seems like a pretty straightforward bug/fix though.
## Changelog
### Fixed
Fixed window centering on high-dpi displays.
# Objective
Fixes#5687
## Solution
Update the methods on the `Entity` struct to be `const`, so we can
define compile-time constants and more generally use them in a const
context.
---
## Changelog
### Added
- Most `Entity` methods are now `const fn`.
# Objective
It's not obvious that the `AssetServerSettings` resource must be added before the `AssetPlugin`.
## Solution
Add a doc comment to this effect.
# Objective
- Reduce debugging burden when using events by telling user when they missed an event.
## Solution
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
- Fixes#5817.
- Removes std::vec::Vec ambiguities in derive_bundle macro
## Solution
Prepend :: to standard library full Vec qualified type name (::std::vec::Vec)
# Objective
- Document `QueryCombinationIter`
## Solution
- Describe the item, add usage and examples
- Copy notes about the number of query items generated from the corresponding query methods (they will be removed in #5742 ([motivation]))
## Additional notes
- Derived from #4989
[motivation]: https://github.com/bevyengine/bevy/pull/4989#issuecomment-1208421496
# Objective
Right now, users have to implement basic system adapters such as `Option` <-> `Result` conversions by themselves. This is slightly annoying and discourages the use of system chaining.
## Solution
Add the module `system_adapter` to the prelude, which contains a collection of common adapters. This is very ergonomic in practice.
## Examples
Convenient early returning.
```rust
use bevy::prelude::*;
App::new()
// If the system fails, just try again next frame.
.add_system(pet_dog.chain(system_adapter::ignore))
.run();
#[derive(Component)]
struct Dog;
fn pet_dog(dogs: Query<(&Name, Option<&Parent>), With<Dog>>) -> Option<()> {
let (dog, dad) = dogs.iter().next()?;
println!("You pet {dog}. He/she/they are a good boy/girl/pupper.");
let (dad, _) = dogs.get(dad?.get()).ok()?;
println!("Their dad's name is {dad}");
Some(())
}
```
Converting the output of a system
```rust
use bevy::prelude::*;
App::new()
.add_system(
find_name
.chain(system_adapter::new(String::from))
.chain(spawn_with_name),
)
.run();
fn find_name() -> &'static str { /* ... */ }
fn spawn_with_name(In(name): In<String>, mut commands: Commands) {
commands.spawn().insert(Name::new(name));
}
```
---
## Changelog
* Added the module `bevy_ecs::prelude::system_adapter`, which contains a collection of common system chaining adapters.
* `new` - Converts a regular fn to a system adapter.
* `unwrap` - Similar to `Result::unwrap`
* `ignore` - Discards the output of the previous system.
# Objective
- Allow users to change the scaling of the UI
- Adopted from #2808
## Solution
- This is an accessibility feature for fixed-size UI elements, allowing the developer to expose a range of UI scales for the player to set a scale that works for their needs.
> - The user can modify the UiScale struct to change the scaling at runtime. This multiplies the Px values by the scale given, while not touching any others.
> - The example showcases how this even allows for fluid transitions
> Here's how the example looks like:
https://user-images.githubusercontent.com/1631166/132979069-044161a9-8e85-45ab-9e93-fcf8e3852c2b.mp4
---
## Changelog
- Added a `UiScale` which can be used to scale all of UI
Co-authored-by: Andreas Weibye <13300393+Weibye@users.noreply.github.com>
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
Help users who are using `load_folder` in wasm builds to find a slightly shorter path to figuring out why their stuff is broken.
## Solution
Adds a warning to `read_directory` in the `WasmAssetIo`.
This is extremely similar to the warning already emitted a few lines below for `watch_for_changes`.
# Objective
Some of the reflection impls for container types had unnecessary `Clone` bounds on their generic arguments. These come from before `FromReflect` when types were instead bound by `Reflect + Clone`. With `FromReflect` this is no longer necessary.
## Solution
Removed all leftover `Clone` bounds from types that use `FromReflect` instead.
## Note
I skipped `Result<T, E>`, `HashSet<T>`, and `Range<T>` since those do not use `FromReflect`. This should probably be handled in a separate PR since it would be a breaking change.
---
## Changelog
- Remove unnecessary `Clone` bounds on reflected containers
# Objective
#5658 made it so that `FromReflect` was used as the bound for `T` in `Option<T>`. However, it did not use this change effectively for the implementation of `Reflect::apply` (it was still using `take`, which would fail for Dynamic types).
Additionally, the changes were not consistent with other methods within the file, such as the ones for `Vec<T>` and `HashMap<K, V>`.
## Solution
Update `Option<T>` to fallback on `FromReflect` if `take` fails, instead of wholly relying on one or the other.
I also chose to update the error messages, as they weren't all too descriptive before.
---
## Changelog
- Use `FromReflect::from_reflect` as a fallback in the `Reflect::apply` implementation for `Option<T>`
# Objective
`SmallVec<T>` was missing a `GetTypeRegistration` impl.
## Solution
Added a `GetTypeRegistration` impl.
---
## Changelog
* Added a `GetTypeRegistration` impl for `SmallVec<T>`
Type registrations were only present for some of the `bevy_math` types, and missing for others. This is a very strange inconsistency, given that they all impl `Reflect` and `FromReflect`. In practice, this means these types cannot be used in scenes.
In particular, this is especially problematic, because `Affine3A` is one of the missing types, and it is now used in `GlobalTransform`. Trying to create a bevy scene that contains `GlobalTransform`s results in an error due to the missing type registration.
# Objective
- While generating https://github.com/jakobhellermann/bevy_reflect_ts_type_export/blob/main/generated/types.ts, I noticed that some types that implement `Reflect` did not register themselves
- `Viewport` isn't reflect but can be (there's a TODO)
## Solution
- register all reflected types
- derive `Reflect` for `Viewport`
## Changelog
- more types are not registered in the type registry
- remove `Serialize`, `Deserialize` impls from `Viewport`
I also decided to remove the `Serialize, Deserialize` from the `Viewport`, since they were (AFAIK) only used for reflection, which now is done without serde. So this is technically a breaking change for people who relied on that impl directly.
Personally I don't think that every bevy type should implement `Serialize, Deserialize`, as that would lead to a ton of code generation that mostly isn't necessary because we can do the same with `Reflect`, but if this is deemed controversial I can remove it from this PR.
## Migration Guide
- `KeyCode` now implements `Reflect` not as `reflect_value`, but with proper struct reflection. The `Serialize` and `Deserialize` impls were removed, now that they are no longer required for scene serialization.
# Objective
Fix a nasty system ordering bug between `update_frusta` and `camera_system` that lead to incorrect frustum s, leading to excessive culling and extremely hard-to-debug visual glitches
## Solution
- add explicit system ordering
# Objective
Very small convenience constructors added to `Size`.
Does not change current examples too much but I'm working on a rather complex UI use-case where this cuts down on some extra typing :)
# Objective
- Fixes#4451
## Solution
- Conditionally compile entity ID cursor as `AtomicI32` when compiling on a platform that does not support 64-bit atomics.
- This effectively raises the MSRV to 1.60 as it uses a `#[cfg]` that was only just stabilized there. (should this be noted in changelog?)
---
## Changelog
- Added `bevy_ecs` support for platforms without 64-bit atomic ints
## Migration Guide
N/A
# Objective
- Morten Mikkelsen clarified that the world normal and tangent must be normalized in the vertex stage and the interpolated values must not be normalized in the fragment stage. This is in order to match the mikktspace approach exactly.
- Fixes#5514 by ensuring the tangent basis matrix (TBN) is orthonormal
## Solution
- Normalize the world normal in the vertex stage and not the fragment stage
- Normalize the world tangent xyz in the vertex stage
- Take into account the sign of the determinant of the local to world matrix when calculating the bitangent
---
## Changelog
- Fixed - scaling a model that uses normal mapping now has correct lighting again
# Objective
- Fixes#5365
- The `assert!()` when the resource from `World::resource_scope` is inserted into the world is not descriptive.
## Solution
- Add more context to the assert inside of `World::resource_scope` when the `FnOnce` param inserts the resource.
# Objective
`FromReflect` is a commonly used component to the Reflect API. It's required as a bound for reflecting things like `Vec<T>` and `HashMap<K, V>` and is generally useful (if not necessary) to derive on most structs or enums.
Currently, however, it is not exported in `bevy_reflect`'s prelude. This means a module that uses `bevy_reflect` might have the following two lines:
```rust
use bevy_reflect::prelude::*;
use bevy_reflect::FromReflect;
```
Additionally, users of the full engine might need to put:
```rust
use bevy::prelude::*;
use bevy::reflect::FromReflect;
```
## Solution
Add `FromReflect` to the prelude of `bevy_reflect`.
---
## Changelog
- Added `FromReflect` to the prelude of `bevy_reflect`
# Objective
- Similar to `SystemChangeTick`, probably somewhat useful for debugging messages.
---
## Changelog
- Added `SystemName` which copies the `SystemMeta::name` field so it can be accessed within a system.
Probably a copy-paste error, but `Add<Color>` and `AddAssign<Color>` should use `rhs.as_hlsa_f32()` instead of `rhs.as_linear_rgba_f32()` when the LHS is a `Color::Hsla`. Fixes#5543.
Co-authored-by: Verte <105466627+vertesians@users.noreply.github.com>
# Objective
The reflection impls on `Option<T>` have the bound `T: Reflect + Clone`. This means that using `FromReflect` requires `Clone` even though we can normally get away with just `FromReflect`.
## Solution
Update the bounds on `Option<T>` to match that of `Vec<T>`, where `T: FromReflect`.
This helps remove a `Clone` implementation that may be undesired but added for the sole purpose of getting the code to compile.
---
## Changelog
* Reflection on `Option<T>` now has `T` bound by `FromReflect` rather than `Reflect + Clone`
* Added a `FromReflect` impl for `Instant`
## Migration Guide
If using `Option<T>` with Bevy's reflection API, `T` now needs to implement `FromReflect` rather than just `Clone`. This can be achieved easily by simply deriving `FromReflect`:
```rust
// OLD
#[derive(Reflect, Clone)]
struct Foo;
let reflected: Box<dyn Reflect> = Box::new(Some(Foo));
// NEW
#[derive(Reflect, FromReflect)]
struct Foo;
let reflected: Box<dyn Reflect> = Box::new(Some(Foo));
```
> Note: You can still derive `Clone`, but it's not required in order to compile.
# Objective
Rust 1.63 resolved [an issue](https://github.com/rust-lang/rust/issues/83701) that prevents you from combining explicit generic arguments with `impl Trait` arguments.
Now, we no longer need to use dynamic dispatch to work around this.
## Migration Guide
The methods `Schedule::get_stage` and `get_stage_mut` now accept `impl StageLabel` instead of `&dyn StageLabel`.
### Before
```rust
let stage = schedule.get_stage_mut::<SystemStage>(&MyLabel)?;
```
### After
```rust
let stage = schedule.get_stage_mut::<SystemStage>(MyLabel)?;
```
# Objective
This PR changes it possible to use vertex colors without a texture using the bevy_sprite ColorMaterial.
Fixes#5679
## Solution
- Made multiplication of the output color independent of the COLOR_MATERIAL_FLAGS_TEXTURE_BIT bit
- Extended mesh2d_vertex_color_texture example to show off both vertex colors and tinting
Not sure if extending the existing example was the right call but it seems to be reasonable to me.
I couldn't find any tests for the shaders and I think adding shader testing would be beyond the scope of this PR. So no tests in this PR. 😬
Co-authored-by: Jonas Wagner <jonas@29a.ch>
# Objective
While poking around the hierarchy code, I wondered why some asserts in tests were duplicated.
Some git blame later, I found out that commit ( 8eb0440f1e ) added already existing asserts while removing others.
## Solution
Remove the duplicated asserts.
# Objective
When an invalid attribute is inserted and the LogPlugin is not enabled the full error is not printed which means makes it hard to diagnose.
## Solution
- Always print the full message in the panic.
## Notes
I originally had a separate error log because I wanted to make it clearer for users, but this is probably causing more issues than necessary.
# Objective
Remove unused `enum DepthCalculation` and its usages. This was used to compute visible entities in the [old renderer](db665b96c0/crates/bevy_render/src/camera/visible_entities.rs), but is now unused.
## Solution
`sed 's/DepthCalculation//g'`
---
## Changelog
### Changed
Removed `bevy_render:📷:DepthCalculation`.
## Migration Guide
Remove references to `bevy_render:📷:DepthCalculation`, such as `use bevy_render:📷:DepthCalculation`. Remove `depth_calculation` fields from Projections.
# Objective
Make CI pass on bevy main.
Update to rust-1.63, updated clippy to 1.63 which introduced the following enhancements:
- [undocumented_unsafe_blocks](https://rust-lang.github.io/rust-clippy/master/index.html#undocumented_unsafe_blocks): Now also lints on unsafe trait implementations
This caught two incorrectly written ( but existing) safety comments for unsafe traits.
## Solution
Fix the comment to use `SAFETY:`
# Objective
While trying out the lint `unsafe_op_in_unsafe_fn` I noticed that `insert_resource_by_id` didn't drop the old value if it already existed, and reimplemented `Column::replace` manually for no apparent reason.
## Solution
- use `Column::replace` and add a test expecting the correct drop count
---
## Changelog
- `World::insert_resource_by_id` will now correctly drop the old resource value, if one already existed
# Objective
- The `Display` impl for `ReflectPathError` is pretty unspecific (e.g. `the current struct doesn't have a field with the given name`
- it has info for better messages available
## Solution
- make the display impl more descriptive by including values from the type
# Objective
- `ReflectMut` served no purpose that wasn't met by `Mut<dyn Reflect>` which is easier to understand since you have to deal with fewer types
- there is another `ReflectMut` type that could be confused with this one
## Solution/Changelog
- relax `T: ?Sized` bound in `Mut<T>`
- replace all instances of `ReflectMut` with `Mut<dyn Reflect>`
# Objective
Provide a safe API to access an `EntityMut`'s `World`.
## Solution
* Add `EntityMut::into_world_mut` for safe access to the entity's world.
---
## Changelog
* Add `EntityMut::into_world_mut` for safe access to the entity's world.
# Objective
`ShaderData` is marked as public, but is an internal type only used by one other
internal type, so it should be made private.
## Solution
`ShaderData` is only used in `ShaderCache`, and the latter is private,
so there is no need to make the former public. This change removes the
`pub` keyword from `ShaderData`, hidding it as the implementation detail
it is.
Split from #5600
# Objective
- I often have UI nodes that are completely transparent and just for organisation
- Don't render them
- I doesn't bring a lot of improvements, but it doesn't add a lot of complexity either
*This PR description is an edited copy of #5007, written by @alice-i-cecile.*
# Objective
Follow-up to https://github.com/bevyengine/bevy/pull/2254. The `Resource` trait currently has a blanket implementation for all types that meet its bounds.
While ergonomic, this results in several drawbacks:
* it is possible to make confusing, silent mistakes such as inserting a function pointer (Foo) rather than a value (Foo::Bar) as a resource
* it is challenging to discover if a type is intended to be used as a resource
* we cannot later add customization options (see the [RFC](https://github.com/bevyengine/rfcs/blob/main/rfcs/27-derive-component.md) for the equivalent choice for Component).
* dependencies can use the same Rust type as a resource in invisibly conflicting ways
* raw Rust types used as resources cannot preserve privacy appropriately, as anyone able to access that type can read and write to internal values
* we cannot capture a definitive list of possible resources to display to users in an editor
## Notes to reviewers
* Review this commit-by-commit; there's effectively no back-tracking and there's a lot of churn in some of these commits.
*ira: My commits are not as well organized :')*
* I've relaxed the bound on Local to Send + Sync + 'static: I don't think these concerns apply there, so this can keep things simple. Storing e.g. a u32 in a Local is fine, because there's a variable name attached explaining what it does.
* I think this is a bad place for the Resource trait to live, but I've left it in place to make reviewing easier. IMO that's best tackled with https://github.com/bevyengine/bevy/issues/4981.
## Changelog
`Resource` is no longer automatically implemented for all matching types. Instead, use the new `#[derive(Resource)]` macro.
## Migration Guide
Add `#[derive(Resource)]` to all types you are using as a resource.
If you are using a third party type as a resource, wrap it in a tuple struct to bypass orphan rules. Consider deriving `Deref` and `DerefMut` to improve ergonomics.
`ClearColor` no longer implements `Component`. Using `ClearColor` as a component in 0.8 did nothing.
Use the `ClearColorConfig` in the `Camera3d` and `Camera2d` components instead.
Co-authored-by: Alice <alice.i.cecile@gmail.com>
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
Co-authored-by: devil-ira <justthecooldude@gmail.com>
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
- I wanted to have controls independent from keyboard layout and found that bevy doesn't have a proper implementation for that
## Solution
- I created a `ScanCode` enum with two hundreds scan codes and updated `keyboard_input_system` to include and update `ResMut<Input<ScanCode>>`
- closes both https://github.com/bevyengine/bevy/issues/2052 and https://github.com/bevyengine/bevy/issues/862
Co-authored-by: Bleb1k <91003089+Bleb1k@users.noreply.github.com>
#4197 intended to remove all `pub` constructors of `Children` and `Parent` and it seems like this one was missed.
Co-authored-by: devil-ira <justthecooldude@gmail.com>
# Objective
- Fixes#5544
- Part of the splitting process of #3692.
## Solution
- Document everything in the `gamepad.rs` file.
- Add a doc example for mocking gamepad input.
---
## Changelog
- Added and updated the documentation inside of the `gamepad.rs` file.
# Objective
- Similar to #5512 , the `View` struct definition in the shaders in `bevy_sprite` and `bevy_ui` were out of sync with the rust-side `ViewUniform`. Only `view_proj` was being used and is the first member and as those shaders are not customisable it makes little difference in practice, unlike for `Mesh2d`.
## Solution
- Sync shader `View` struct definition in `bevy_sprite` and `bevy_ui` with the correct definition that matches `ViewUniform`
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
# Objective
Change frametimediagnostic from seconds to milliseconds because this will always be less than one seconds and is the common diagnostic display unit for game engines.
## Solution
- multiplied the existing value by 1000
---
## Changelog
Frametimes are now reported in milliseconds
Co-authored-by: Syama Mishra <38512086+SyamaMishra@users.noreply.github.com>
Co-authored-by: McSpidey <mcspidey@gmail.com>
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
