# Objective
- Contributes to #16877
## Solution
- Initial creation of `bevy_platform_support` crate.
- Moved `bevy_utils::Instant` into new `bevy_platform_support` crate.
- Moved `portable-atomic`, `portable-atomic-util`, and
`critical-section` into new `bevy_platform_support` crate.
## Testing
- CI
---
## Showcase
Instead of needing code like this to import an `Arc`:
```rust
#[cfg(feature = "portable-atomic")]
use portable_atomic_util::Arc;
#[cfg(not(feature = "portable-atomic"))]
use alloc::sync::Arc;
```
We can now use:
```rust
use bevy_platform_support::sync::Arc;
```
This applies to many other types, but the goal is overall the same:
allowing crates to use `std`-like types without the boilerplate of
conditional compilation and platform-dependencies.
## Migration Guide
- Replace imports of `bevy_utils::Instant` with
`bevy_platform_support::time::Instant`
- Replace imports of `bevy::utils::Instant` with
`bevy::platform_support::time::Instant`
## Notes
- `bevy_platform_support` hasn't been reserved on `crates.io`
- ~~`bevy_platform_support` is not re-exported from `bevy` at this time.
It may be worthwhile exporting this crate, but I am unsure of a
reasonable name to export it under (`platform_support` may be a bit
wordy for user-facing).~~
- I've included an implementation of `Instant` which is suitable for
`no_std` platforms that are not Wasm for the sake of eliminating feature
gates around its use. It may be a controversial inclusion, so I'm happy
to remove it if required.
- There are many other items (`spin`, `bevy_utils::Sync(Unsafe)Cell`,
etc.) which should be added to this crate. I have kept the initial scope
small to demonstrate utility without making this too unwieldy.
---------
Co-authored-by: TimJentzsch <TimJentzsch@users.noreply.github.com>
Co-authored-by: Chris Russell <8494645+chescock@users.noreply.github.com>
Co-authored-by: François Mockers <francois.mockers@vleue.com>
# Objective
Fixes#17164
## Solution
- Switch to `ubuntu-latest`
- Remove `libegl1-mesa`. It seems that whatever this was providing is
now covered by other deps or their deps?
## Testing
Tested [over here](https://github.com/rparrett/bevy/pull/21) by
commenting out workflow `if` condition.
Makes use of `std` explicit, simplifying a possible `no_std` port.
# Objective
- Contributes to #15460
- Simplify future `no_std` work on `bevy_asset`
## Solution
- Add `#![no_std]` to switch to `core::prelude` instead of
`std::prelude`
## Testing
- CI
---
## Notes
This is entirely a change around the names of imports and has no impact
on functionality. This just reduces the quantity of changes involved in
the (likely more controversial) `no_std`-ification of `bevy_asset`.
Fixes#17397.
Also renamed all variants into present-tense.
## Migration Guide
- `PointerAction::Pressed` has been seperated into two variants,
`PointerAction::Press` and `PointerAction::Release`.
- `PointerAction::Moved` has been renamed to `PointerAction::Move`.
- `PointerAction::Canceled` has been renamed to `PointerAction::Cancel`.
This commit overhauls the documentation in the Bevy scene loading
example. It adds thorough explanatory comments to guide new Rust and
Bevy developers. The rewritten docs clarify how to:
- Register types for reflection, enabling serialization and dynamic
property access
- Skip serializing certain fields with `#[reflect(skip_serializing)]`
- Use `FromWorld` for components that require runtime initialization
- Store and serialize `Resources` in scene files
- Load scenes using a `DynamicSceneRoot` and handle updates in a system
- Serialize a brand-new scene to a separate file asynchronously using
`IoTaskPool`
These additions aim to provide a clear, step-by-step reference that
demonstrates how to implement a scene-based workflow, making it easier
for beginners and experienced developers alike to use Bevy’s scene
system effectively.
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
# Objective
It's not immediately obvious that `TargetCamera` only works with UI node
entities. It's natural to assume from looking at something like the
`multiple_windows` example that it will work with everything.
## Solution
Rename `TargetCamera` to `UiTargetCamera`.
## Migration Guide
`TargetCamera` has been renamed to `UiTargetCamera`.
# Objective
Segment2d and Segment3d are currently hard to work with because unlike
many other primary shapes, they are bound to the origin.
The objective of this PR is to allow these segments to exist anywhere in
cartesian space, making them much more useful in a variety of contexts.
## Solution
Reworking the existing segment type's internal fields and methods to
allow them to exist anywhere in cartesian space.
I have done both reworks for 2d and 3d segments but I was unsure if I
should just have it all here or not so feel free to tell me how I should
proceed, for now I have only pushed Segment2d changes.
As I am not a very seasoned contributor, this first implementation is
very likely sloppy and will need some additional work from my end, I am
open to all criticisms and willing to work to get this to bevy's
standards.
## Testing
I am not very familiar with the standards of testing. Of course my
changes had to pass the thorough existing tests for primitive shapes.
I also checked the gizmo 2d shapes intersection example and everything
looked fine.
I did add a few utility methods to the types that have no tests yet. I
am willing to implement some if it is deemed necessary
## Migration Guide
The segment type constructors changed so if someone previously created a
Segment2d with a direction and length they would now need to use the
`from_direction` constructor
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
Co-authored-by: Joona Aalto <jondolf.dev@gmail.com>
This adds support for one-to-many non-fragmenting relationships (with
planned paths for fragmenting and non-fragmenting many-to-many
relationships). "Non-fragmenting" means that entities with the same
relationship type, but different relationship targets, are not forced
into separate tables (which would cause "table fragmentation").
Functionally, this fills a similar niche as the current Parent/Children
system. The biggest differences are:
1. Relationships have simpler internals and significantly improved
performance and UX. Commands and specialized APIs are no longer
necessary to keep everything in sync. Just spawn entities with the
relationship components you want and everything "just works".
2. Relationships are generalized. Bevy can provide additional built in
relationships, and users can define their own.
**REQUEST TO REVIEWERS**: _please don't leave top level comments and
instead comment on specific lines of code. That way we can take
advantage of threaded discussions. Also dont leave comments simply
pointing out CI failures as I can read those just fine._
## Built on top of what we have
Relationships are implemented on top of the Bevy ECS features we already
have: components, immutability, and hooks. This makes them immediately
compatible with all of our existing (and future) APIs for querying,
spawning, removing, scenes, reflection, etc. The fewer specialized APIs
we need to build, maintain, and teach, the better.
## Why focus on one-to-many non-fragmenting first?
1. This allows us to improve Parent/Children relationships immediately,
in a way that is reasonably uncontroversial. Switching our hierarchy to
fragmenting relationships would have significant performance
implications. ~~Flecs is heavily considering a switch to non-fragmenting
relations after careful considerations of the performance tradeoffs.~~
_(Correction from @SanderMertens: Flecs is implementing non-fragmenting
storage specialized for asset hierarchies, where asset hierarchies are
many instances of small trees that have a well defined structure)_
2. Adding generalized one-to-many relationships is currently a priority
for the [Next Generation Scene / UI
effort](https://github.com/bevyengine/bevy/discussions/14437).
Specifically, we're interested in building reactions and observers on
top.
## The changes
This PR does the following:
1. Adds a generic one-to-many Relationship system
3. Ports the existing Parent/Children system to Relationships, which now
lives in `bevy_ecs::hierarchy`. The old `bevy_hierarchy` crate has been
removed.
4. Adds on_despawn component hooks
5. Relationships can opt-in to "despawn descendants" behavior, meaning
that the entire relationship hierarchy is despawned when
`entity.despawn()` is called. The built in Parent/Children hierarchies
enable this behavior, and `entity.despawn_recursive()` has been removed.
6. `world.spawn` now applies commands after spawning. This ensures that
relationship bookkeeping happens immediately and removes the need to
manually flush. This is in line with the equivalent behaviors recently
added to the other APIs (ex: insert).
7. Removes the ValidParentCheckPlugin (system-driven / poll based) in
favor of a `validate_parent_has_component` hook.
## Using Relationships
The `Relationship` trait looks like this:
```rust
pub trait Relationship: Component + Sized {
type RelationshipSources: RelationshipSources<Relationship = Self>;
fn get(&self) -> Entity;
fn from(entity: Entity) -> Self;
}
```
A relationship is a component that:
1. Is a simple wrapper over a "target" Entity.
2. Has a corresponding `RelationshipSources` component, which is a
simple wrapper over a collection of entities. Every "target entity"
targeted by a "source entity" with a `Relationship` has a
`RelationshipSources` component, which contains every "source entity"
that targets it.
For example, the `Parent` component (as it currently exists in Bevy) is
the `Relationship` component and the entity containing the Parent is the
"source entity". The entity _inside_ the `Parent(Entity)` component is
the "target entity". And that target entity has a `Children` component
(which implements `RelationshipSources`).
In practice, the Parent/Children relationship looks like this:
```rust
#[derive(Relationship)]
#[relationship(relationship_sources = Children)]
pub struct Parent(pub Entity);
#[derive(RelationshipSources)]
#[relationship_sources(relationship = Parent)]
pub struct Children(Vec<Entity>);
```
The Relationship and RelationshipSources derives automatically implement
Component with the relevant configuration (namely, the hooks necessary
to keep everything in sync).
The most direct way to add relationships is to spawn entities with
relationship components:
```rust
let a = world.spawn_empty().id();
let b = world.spawn(Parent(a)).id();
assert_eq!(world.entity(a).get::<Children>().unwrap(), &[b]);
```
There are also convenience APIs for spawning more than one entity with
the same relationship:
```rust
world.spawn_empty().with_related::<Children>(|s| {
s.spawn_empty();
s.spawn_empty();
})
```
The existing `with_children` API is now a simpler wrapper over
`with_related`. This makes this change largely non-breaking for existing
spawn patterns.
```rust
world.spawn_empty().with_children(|s| {
s.spawn_empty();
s.spawn_empty();
})
```
There are also other relationship APIs, such as `add_related` and
`despawn_related`.
## Automatic recursive despawn via the new on_despawn hook
`RelationshipSources` can opt-in to "despawn descendants" behavior,
which will despawn all related entities in the relationship hierarchy:
```rust
#[derive(RelationshipSources)]
#[relationship_sources(relationship = Parent, despawn_descendants)]
pub struct Children(Vec<Entity>);
```
This means that `entity.despawn_recursive()` is no longer required.
Instead, just use `entity.despawn()` and the relevant related entities
will also be despawned.
To despawn an entity _without_ despawning its parent/child descendants,
you should remove the `Children` component first, which will also remove
the related `Parent` components:
```rust
entity
.remove::<Children>()
.despawn()
```
This builds on the on_despawn hook introduced in this PR, which is fired
when an entity is despawned (before other hooks).
## Relationships are the source of truth
`Relationship` is the _single_ source of truth component.
`RelationshipSources` is merely a reflection of what all the
`Relationship` components say. By embracing this, we are able to
significantly improve the performance of the system as a whole. We can
rely on component lifecycles to protect us against duplicates, rather
than needing to scan at runtime to ensure entities don't already exist
(which results in quadratic runtime). A single source of truth gives us
constant-time inserts. This does mean that we cannot directly spawn
populated `Children` components (or directly add or remove entities from
those components). I personally think this is a worthwhile tradeoff,
both because it makes the performance much better _and_ because it means
theres exactly one way to do things (which is a philosophy we try to
employ for Bevy APIs).
As an aside: treating both sides of the relationship as "equivalent
source of truth relations" does enable building simple and flexible
many-to-many relationships. But this introduces an _inherent_ need to
scan (or hash) to protect against duplicates.
[`evergreen_relations`](https://github.com/EvergreenNest/evergreen_relations)
has a very nice implementation of the "symmetrical many-to-many"
approach. Unfortunately I think the performance issues inherent to that
approach make it a poor choice for Bevy's default relationship system.
## Followup Work
* Discuss renaming `Parent` to `ChildOf`. I refrained from doing that in
this PR to keep the diff reasonable, but I'm personally biased toward
this change (and using that naming pattern generally for relationships).
* [Improved spawning
ergonomics](https://github.com/bevyengine/bevy/discussions/16920)
* Consider adding relationship observers/triggers for "relationship
targets" whenever a source is added or removed. This would replace the
current "hierarchy events" system, which is unused upstream but may have
existing users downstream. I think triggers are the better fit for this
than a buffered event queue, and would prefer not to add that back.
* Fragmenting relations: My current idea hinges on the introduction of
"value components" (aka: components whose type _and_ value determines
their ComponentId, via something like Hashing / PartialEq). By labeling
a Relationship component such as `ChildOf(Entity)` as a "value
component", `ChildOf(e1)` and `ChildOf(e2)` would be considered
"different components". This makes the transition between fragmenting
and non-fragmenting a single flag, and everything else continues to work
as expected.
* Many-to-many support
* Non-fragmenting: We can expand Relationship to be a list of entities
instead of a single entity. I have largely already written the code for
this.
* Fragmenting: With the "value component" impl mentioned above, we get
many-to-many support "for free", as it would allow inserting multiple
copies of a Relationship component with different target entities.
Fixes#3742 (If this PR is merged, I think we should open more targeted
followup issues for the work above, with a fresh tracking issue free of
the large amount of less-directed historical context)
Fixes#17301Fixes#12235Fixes#15299Fixes#15308
## Migration Guide
* Replace `ChildBuilder` with `ChildSpawnerCommands`.
* Replace calls to `.set_parent(parent_id)` with
`.insert(Parent(parent_id))`.
* Replace calls to `.replace_children()` with `.remove::<Children>()`
followed by `.add_children()`. Note that you'll need to manually despawn
any children that are not carried over.
* Replace calls to `.despawn_recursive()` with `.despawn()`.
* Replace calls to `.despawn_descendants()` with
`.despawn_related::<Children>()`.
* If you have any calls to `.despawn()` which depend on the children
being preserved, you'll need to remove the `Children` component first.
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
Added an external assets section to .gitignore. This prevents
contributors from accidentally adding or committing them.
I believe currently the only externel asset is the meshlet bunny.
This allows this:
```rust
#[derive(Component)]
#[require(StateScoped<MyState>(StateScoped(MyState)))]
struct ComponentA;
```
To be shortened to this:
```rust
#[derive(Component)]
#[require(StateScoped<MyState>)]
struct ComponentA;
```
When `MyState` implements `Default`.
Signed-off-by: Jean Mertz <git@jeanmertz.com>
# Objective
Occasionally bevy users will want to store systems or observer systems
in a component or resource, but they first try to store `IntoSystem`
instead of `System`, which leads to some headaches having to deal with
the `M` marker type parameter. We should recommend they use the `X`
trait instead of the `IntoX` trait in that case, as well for returning
from a function.
## Solution
Add usage notes to the `IntoX` traits about using `X` instead.
# Objective
While working on more complex directional navigation work, I noticed a
few small things.
## Solution
Rather than stick them in a bigger PR, split them out now.
- Include more useful information when responding to
`DirectionalNavigationError`.
- Use the less controversial `Click` events (rather than `Pressed`) in
the example
- Implement add_looping_edges in terms of `add_edges`. Thanks @rparrett
for the idea.
## Testing
Ran the `directional_navigation` example and things still work.
# Objective
The safety documentation for `Ptr::assert_unique` is incomplete.
Currently it only mentions the existence of other `Ptr` instances, but
it should also mention that the underlying data must be mutable and that
there cannot be active references to it.
# Objective
While `add_looping_edges` is a helpful method for manually defining
directional navigation maps, we don't always want to loop around!
## Solution
Add a non-looping variant.
These commits are cherrypicked from the more complex #17247.
## Testing
I've updated the `directional_navigation` example to use these changes,
and verified that it works.
---------
Co-authored-by: Rob Parrett <robparrett@gmail.com>
Co-authored-by: Benjamin Brienen <benjamin.brienen@outlook.com>
# Objective
- Fix issue identified on the [Discord
server](https://discord.com/channels/691052431525675048/691052431974465548/1328922812530036839)
## Solution
- Implement `Clone` for `QueryIter` using the existing
`QueryIter::remaining` method
## Testing
- CI
---
## Showcase
Users can now explicitly clone a read-only `QueryIter`:
```rust
fn combinations(query: Query<&ComponentA>) {
let mut iter = query.iter();
while let Some(a) = iter.next() {
// Can now clone rather than use remaining
for b in iter.clone() {
// Check every combination (a, b)
}
}
}
```
## Notes
This doesn't add any new functionality outside the context of generic
code (e.g., `T: Iterator<...> + Clone`), it's mostly for
discoverability. Users are more likely to be familiar with
`Clone::clone` than they are with the methods on `QueryIter`.
# Objective
As raised in https://github.com/bevyengine/bevy/pull/17317, the `Event:
Component` trait bound is confusing to users.
In general, a type `E` (like `AppExit`) which implements `Event` should
not:
- be stored as a component on an entity
- be a valid option for `Query<&AppExit>`
- require the storage type and other component metadata to be specified
Events are not components (even if they one day use some of the same
internal mechanisms), and this trait bound is confusing to users.
We're also automatically generating `Component` impls with our derive
macro, which should be avoided when possible to improve explicitness and
avoid conflicts with user impls.
Closes#17317, closes#17333
## Solution
- We only care that each unique event type gets a unique `ComponentId`
- dynamic events need their own tools for getting identifiers anyways
- This avoids complicating the internals of `ComponentId` generation.
- Clearly document why this cludge-y solution exists.
In the medium term, I think that either a) properly generalizing
`ComponentId` (and moving it into `bevy_reflect?) or b) using a
new-typed `Entity` as the key for events is more correct. This change is
stupid simple though, and removes the offending trait bound in a way
that doesn't introduce complex tech debt and does not risk changes to
the internals.
This change does not:
- restrict our ability to implement dynamic buffered events (the main
improvement over #17317)
- there's still a fair bit of work to do, but this is a step in the
right direction
- limit our ability to store event metadata on entities in the future
- make it harder for users to work with types that are both events and
components (just add the derive / trait bound)
## Migration Guide
The `Event` trait no longer requires the `Component` trait. If you were
relying on this behavior, change your trait bounds from `Event` to
`Event + Component`. If you also want your `Event` type to implement
`Component`, add a derive.
---------
Co-authored-by: Chris Russell <8494645+chescock@users.noreply.github.com>
# Objective
UI node Outlines are clipped using their parent's clipping rect instead
of their own.
## Solution
Clip outlines using the UI node's own clipping rect.
I noticed that this component was not being returned correctly by the
`bevy_remote` api
```json
"errors": {
"bevy_picking::focus::PickingInteraction": {
"code": -23402,
"message": "Unknown component type: `bevy_picking::focus::PickingInteraction`"
}
}
```
# Objective
- Follow up work from
https://github.com/bevyengine/bevy/pull/17121#issuecomment-2576615700 to
keep the `cursor.rs` file more manageable.
## Solution
- Move `CustomCursor` and make it compile.
## Testing
- Ran the example: `cargo run --example custom_cursor_image
--features=custom_cursor`
- CI
# Objective
Fixes https://github.com/bevyengine/bevy/issues/17111
## Solution
Move `#![warn(clippy::allow_attributes,
clippy::allow_attributes_without_reason)]` to the workspace `Cargo.toml`
## Testing
Lots of CI testing, and local testing too.
---------
Co-authored-by: Benjamin Brienen <benjamin.brienen@outlook.com>
# Objective
- Bevy 0.15 added support for custom cursor images in
https://github.com/bevyengine/bevy/pull/14284.
- However, to do animated cursors using the initial support shipped in
0.15 means you'd have to animate the `Handle<Image>`: You can't use a
`TextureAtlas` like you can with sprites and UI images.
- For my use case, my cursors are spritesheets. To animate them, I'd
have to break them down into multiple `Image` assets, but that seems
less than ideal.
## Solution
- Allow users to specify a `TextureAtlas` field when creating a custom
cursor image.
- To create parity with Bevy's `TextureAtlas` support on `Sprite`s and
`ImageNode`s, this also allows users to specify `rect`, `flip_x` and
`flip_y`. In fact, for my own use case, I need to `flip_y`.
## Testing
- I added unit tests for `calculate_effective_rect` and
`extract_and_transform_rgba_pixels`.
- I added a brand new example for custom cursor images. It has controls
to toggle fields on and off. I opted to add a new example because the
existing cursor example (`window_settings`) would be far too messy for
showcasing these custom cursor features (I did start down that path but
decided to stop and make a brand new example).
- The new example uses a [Kenny cursor icon] sprite sheet. I included
the licence even though it's not required (and it's CC0).
- I decided to make the example just loop through all cursor icons for
its animation even though it's not a _realistic_ in-game animation
sequence.
- I ran the PNG through https://tinypng.com. Looks like it's about 35KB.
- I'm open to adjusting the example spritesheet if required, but if it's
fine as is, great.
[Kenny cursor icon]: https://kenney-assets.itch.io/crosshair-pack
---
## Showcase
https://github.com/user-attachments/assets/8f6be8d7-d1d4-42f9-b769-ef8532367749
## Migration Guide
The `CustomCursor::Image` enum variant has some new fields. Update your
code to set them.
Before:
```rust
CustomCursor::Image {
handle: asset_server.load("branding/icon.png"),
hotspot: (128, 128),
}
```
After:
```rust
CustomCursor::Image {
handle: asset_server.load("branding/icon.png"),
texture_atlas: None,
flip_x: false,
flip_y: false,
rect: None,
hotspot: (128, 128),
}
```
## References
- Feature request [originally raised in Discord].
[originally raised in Discord]:
https://discord.com/channels/691052431525675048/692572690833473578/1319836362219847681
# Objective
Support the parametrization of the WS_CLIPCHILDREN style on Windows.
Fixes#16544
## Solution
Added a window configuration in bevy_winit to control the usage of the
WS_CLIPCHILDREN style.
## Testing
- Did you test these changes? If so, how?
I did. I was able to create a Wry Webview with a transparent HTML
document and was also able to see my Bevy scene behind the webview
elements.
- Are there any parts that need more testing?
I don't believe so. I assume the option is extensively tested within
winit itself.
- How can other people (reviewers) test your changes? Is there anything
specific they need to know?
Test repositiory [here](https://github.com/nicholasc/bevy_wry_test).
Bevy's path will need to be updated in the Cargo.toml
- If relevant, what platforms did you test these changes on, and are
there any important ones you can't test?
This is a Windows specific issue. Should be tested accordingly.
---------
Co-authored-by: jf908 <jf908@users.noreply.github.com>
# Objective
- Currently, the `ObservedBy`-component is only public within the
`bevy_ecs` crate. Sometimes it is desirable to refer to this component
in the "game-code". Two examples that come in mind:
- Clearing all components in an entity, but intending to keep the
existing observers: Making `ObservedBy` public allows us to use
`commands.entity(entity).retain::<ObservedBy>();`, which clears all
other components, but keeps `ObservedBy`, which prevents the Observers
from despawning.
- The opposite of the above, clearing all of entities' Observers:
`commands.entity(entity).remove::<ObservedBy>` will despawn all
associated Observers. Admittedly, a cleaner solution would be something
like `commands.entity(entity).clear_observers()`, but this is
sufficient.
## Solution
- Removed `(crate)` "rule" and added `ObservedBy` to the prelude-module
## Testing
- Linked `bevy_ecs` locally with another project to see if `ObservedBy`
could be referenced.
# Objective
resolves#17326.
## Solution
Simply added the suggested run condition.
## Testing
A self-explanatory run condition. Fully verified by the operation of
`QueryFilter` in a system.
# Objective
- https://github.com/bevyengine/bevy/issues/17111
## Solution
Set the `clippy::allow_attributes` and
`clippy::allow_attributes_without_reason` lints to `warn`, and bring
`bevy_ecs` in line with the new restrictions.
## Testing
This PR is a WIP; testing will happen after it's finished.
# Objective
- https://github.com/bevyengine/bevy/issues/17111
## Solution
Set the `clippy::allow_attributes` and
`clippy::allow_attributes_without_reason` lints to `warn`, and bring
`bevy_macro_utils` in line with the new restrictions.
## Testing
`cargo clippy --tests --all-features --package bevy_macro_utils` was
run, and no warnings were encountered.
# Objective
- https://github.com/bevyengine/bevy/issues/17111
## Solution
Set the `clippy::allow_attributes` and
`clippy::allow_attributes_without_reason` lints to `warn`, and bring
`bevy_input_focus` in line with the new restrictions.
## Testing
`cargo clippy --tests --all-features --features
bevy_math/std,bevy_input/smol_str --package bevy_input_focus` was run,
and only an unrelated warning from `bevy_ecs` was encountered.
I could not test without the `bevy_math/std` feature due to compilation
errors with `glam`. Additionally, I had to use the `bevy_input/smol_str`
feature, as it appears some of `bevy_input_focus`' tests rely on that
feature. I will investigate these issues further, and make issues/PRs as
necessary.
# Objective
- https://github.com/bevyengine/bevy/issues/17111
## Solution
Set the `clippy::allow_attributes` and
`clippy::allow_attributes_without_reason` lints to `warn`, and bring
`bevy_dylib` in line with the new restrictions.
## Testing
`cargo clippy --tests --all-features --package bevy_dylib` was run, and
no warnings were encountered.
I would've skipped over this crate if there weren't the two lint
attributes in it - might as well handle it now, y'know?
# Objective
- https://github.com/bevyengine/bevy/issues/17111
## Solution
Set the `clippy::allow_attributes` and
`clippy::allow_attributes_without_reason` lints to `warn`, and bring
`bevy_core_pipeline` in line with the new restrictions.
## Testing
`cargo clippy` and `cargo test --package bevy_core_pipeline` were run,
and no warnings were encountered.
This commit allows Bevy to use `multi_draw_indirect_count` for drawing
meshes. The `multi_draw_indirect_count` feature works just like
`multi_draw_indirect`, but it takes the number of indirect parameters
from a GPU buffer rather than specifying it on the CPU.
Currently, the CPU constructs the list of indirect draw parameters with
the instance count for each batch set to zero, uploads the resulting
buffer to the GPU, and dispatches a compute shader that bumps the
instance count for each mesh that survives culling. Unfortunately, this
is inefficient when we support `multi_draw_indirect_count`. Draw
commands corresponding to meshes for which all instances were culled
will remain present in the list when calling
`multi_draw_indirect_count`, causing overhead. Proper use of
`multi_draw_indirect_count` requires eliminating these empty draw
commands.
To address this inefficiency, this PR makes Bevy fully construct the
indirect draw commands on the GPU instead of on the CPU. Instead of
writing instance counts to the draw command buffer, the mesh
preprocessing shader now writes them to a separate *indirect metadata
buffer*. A second compute dispatch known as the *build indirect
parameters* shader runs after mesh preprocessing and converts the
indirect draw metadata into actual indirect draw commands for the GPU.
The build indirect parameters shader operates on a batch at a time,
rather than an instance at a time, and as such each thread writes only 0
or 1 indirect draw parameters, simplifying the current logic in
`mesh_preprocessing`, which currently has to have special cases for the
first mesh in each batch. The build indirect parameters shader emits
draw commands in a tightly packed manner, enabling maximally efficient
use of `multi_draw_indirect_count`.
Along the way, this patch switches mesh preprocessing to dispatch one
compute invocation per render phase per view, instead of dispatching one
compute invocation per view. This is preparation for two-phase occlusion
culling, in which we will have two mesh preprocessing stages. In that
scenario, the first mesh preprocessing stage must only process opaque
and alpha tested objects, so the work items must be separated into those
that are opaque or alpha tested and those that aren't. Thus this PR
splits out the work items into a separate buffer for each phase. As this
patch rewrites so much of the mesh preprocessing infrastructure, it was
simpler to just fold the change into this patch instead of deferring it
to the forthcoming occlusion culling PR.
Finally, this patch changes mesh preprocessing so that it runs
separately for indexed and non-indexed meshes. This is because draw
commands for indexed and non-indexed meshes have different sizes and
layouts. *The existing code is actually broken for non-indexed meshes*,
as it attempts to overlay the indirect parameters for non-indexed meshes
on top of those for indexed meshes. Consequently, right now the
parameters will be read incorrectly when multiple non-indexed meshes are
multi-drawn together. *This is a bug fix* and, as with the change to
dispatch phases separately noted above, was easiest to include in this
patch as opposed to separately.
## Migration Guide
* Systems that add custom phase items now need to populate the indirect
drawing-related buffers. See the `specialized_mesh_pipeline` example for
an example of how this is done.
# Objective
- While all
[`MeshBuilder`](https://dev-docs.bevyengine.org/bevy/prelude/trait.MeshBuilder.html)s
can be created by first creating the primitive `bevy_math` type and
using
[`Meshable`](https://dev-docs.bevyengine.org/bevy/prelude/trait.Meshable.html),
most builders have their own `const` constructors that can be used
instead. Some builders are missing constructors, however, making them
unavailable in `const` contexts.
## Solution
- Add a `const` constructor for `RegularPolygonMeshBuilder`,
`RhombusMeshBuilder`, `Triangle2dMeshBuilder`, and
`RectangleMeshBuilder`.
- Add a note on the requirements of `ConvexPolygonMeshBuilder`, and
recommend using `ConvexPolygon::new().mesh()` instead.
- A constructor cannot easily be created for this type, since it
requires duplicating all of `ConvexPolygon::new()`'s verification code.
I may be able to work around this, but it requires touching a bit more
code surface. Opinions?
## Testing
Not much beyond CI! The changes are trivial enough that only a cursory
glance for typos and switched variables should be necessary.
## Note for Reviewers
Hi! I haven't directly used the types I modify in this PR beyond some
benchmarking work I did this morning. If you're familiar with these
types, please let me know if any of the constructors need additional
validation (or if the constructors shouldn't be there at all). Thanks!
---------
Co-authored-by: IQuick 143 <IQuick143cz@gmail.com>
# Objective
allow setting ambient light via component on cameras.
arguably fixes#7193
note i chose to use a component rather than an entity since it was not
clear to me how to manage multiple ambient sources for a single
renderlayer, and it makes for a very small changeset.
## Solution
- make ambient light a component as well as a resource
- extract it
- use the component if present on a camera, fallback to the resource
## Testing
i added
```rs
if index == 1 {
commands.entity(camera).insert(AmbientLight{
color: Color::linear_rgba(1.0, 0.0, 0.0, 1.0),
brightness: 1000.0,
..Default::default()
});
}
```
at line 84 of the split_screen example
---------
Co-authored-by: François Mockers <mockersf@gmail.com>
# Objective
Add a method to mutate components with BRP.
Currently the only way to modify a component on an entity with BRP is to
insert a new one with the new values. This isn't ideal for several
reasons, one reason being that the client has to know what all the
fields are of the component and stay in sync with the server.
## Solution
Add a new BRP method called `bevy/mutate_component` to mutate a single
field in a component on an entity.
## Testing
Tested on a simple scene on all `Transform`, `Name`, and a custom
component.
---
## Showcase
Example JSON-RPC request to change the `Name` of an entity to "New
name!"
```json
{
"jsonrpc": "2.0",
"id": 0,
"method": "bevy/mutate_component",
"params": {
"entity": 4294967308,
"component": "bevy_ecs::name::Name",
"path": ".name",
"value": "New name!"
}
}
```
Or setting the X translation to 10.0 on a Transform:
```json
{
"jsonrpc": "2.0",
"id": 0,
"method": "bevy/mutate_component",
"params": {
"entity": 4294967308,
"component": "bevy_transform::components::transform::Transform",
"path": ".translation.x",
"value": 10.0
}
}
```
Clip of my Emacs BRP package using this method:
https://github.com/user-attachments/assets/a786b245-5c20-4189-859f-2261c5086a68
---------
Co-authored-by: François Mockers <mockersf@gmail.com>
# Objective
- Stop bevy from crashing when losing window focus
## Solution
- The InputFocus resource is optional but is accessed unconditionally in
bevy_winit. Make it optional.
## Testing
- Ran the window_settings example
## Note
It's possible this might not be a full fix for the issue, but this stop
bevy from crashing.
Closes#16961Closes#17227
# Objective
Diagnostics is reporting incorrect FPS and frame time.
## Solution
Looks like the smoothing value should be `2 / (history_length + 1)` not
`(history_length + 1) / 2`.
Co-authored-by: François Mockers <mockersf@gmail.com>
# Objective
Building upon https://github.com/bevyengine/bevy/pull/17191, improve the
`animated_mesh` example by removing code, adding comments, and making
the example more c&p'able.
## Solution
- Split the setup function in two to clarify what the example is
demonstrating.
- `setup_mesh_and_animation` is the demonstration.
- `setup_camera_and_environment` just sets up the example app.
- Changed the animation playing to use `AnimationPlayer` directly
instead of creating `AnimationTransitions`.
- This appears sufficient when only playing a single animation.
- Added a comment pointing users to an example of multiple animations.
- Changed the animation to be the run cycle.
- I think it got accidentally changed to the idle in
[#17191](https://github.com/bevyengine/bevy/pull/17191), so this is
reverting back to the original.
- Note that we can improve it to select the animation by name if
[#16529](https://github.com/bevyengine/bevy/pull/16529) lands.
- Renamed `FOX_PATH` to a more neutral `GLTF_PATH`.
- Updated the example descriptions to mention the fox.
- This adds a little character and hints that the example involves
character animation.
- Removed a seemingly redundant `AnimationGraphHandle` component.
- Removed an unnecessary `clone()`.
- Added various comments.
## Notes
- A draft of this PR was discussed on Discord:
https://discord.com/channels/691052431525675048/1326910663972618302/1326920498663133348
- There was discord discussion on whether a component is "inserted
onto", "inserted into" or "added to" an entity.
- "Added to" is most common in code and docs, and seems best to me. But
it awkwardly differs from the name of `EntityCommands::insert`.
- This PR prefers "added to".
- I plan to follow up this PR with similar changes to the
`animated_mesh_control` and `animated_mesh_events` examples.
- But I could roll them into this PR if requested.
## Testing
`cargo run --example animated_mesh`
---------
Co-authored-by: François Mockers <mockersf@gmail.com>
# Objective
Necessary conditions:
* Scale factor != 1
* Text is being displayed with Text2d
* The primary window is closed on a frame where the text or text's
bounds are modified.
Then when `update_text2d_layout` runs, it finds no primary window and
assumes a scale factor of 1.
The previous scale_factor was not equal to 1 and the text pipeline's old
font atlases were created for a non-1 scale factor, so it creates new
font atlases even though the app is closing.
The bug was first identified in #6666
## Minimal Example
```rust
use bevy::prelude::*;
fn main() {
App::new()
.add_plugins(DefaultPlugins.set(WindowPlugin {
primary_window: Some(Window {
present_mode: bevy:🪟:PresentMode::Immediate,
..Default::default()
}),
..default()
}))
.insert_resource(UiScale { scale: std::f64::consts::PI })
.add_startup_system(setup)
.add_system(update)
.run();
}
fn setup(mut commands: Commands, asset_server: Res<AssetServer>) {
commands.spawn(Camera2dBundle::default());
commands.spawn(Text2dBundle {
text: Text {
sections: (0..10).map(|i| TextSection {
value: i.to_string(),
style: TextStyle {
font: asset_server.load("fonts/FiraSans-Bold.ttf"),
font_size: (10 + i) as f32,
color: Color::WHITE,
}
}).collect(),
..Default::default()
},
..Default::default()
});
}
fn update(mut text: Query<&mut Text>) {
for mut text in text.iter_mut() {
text.set_changed();
}
}
```
## Output
On closing the window you'll see the warning (if you don't, increase the
number of text sections):
```
WARN bevy_text::glyph_brush: warning[B0005]: Number of font atlases has exceeded the maximum of 16. Performance and memory usage may suffer.
```
The app should only create font atlases on startup, but it doesn't
display this warning until after you close the window
## Solution
Skip `update_text_layout` when there is no primary window.
## Changelog
* If no primary window is found, skip `update_text2d_layout`.
* Added a `Local` flag `skipped` to `update_text2d_layout`. This should
ensure there are no edge cases where text might not get drawn at all.
---------
Co-authored-by: François Mockers <mockersf@gmail.com>
# Objective
- Use `Clone` on `SystemParam`, when applicable, in a generic context.
## Solution
- Add some derives
## Testing
- I ran `cargo test` once.
- I didn't even look at the output.
---------
Co-authored-by: François Mockers <mockersf@gmail.com>
# Objective
Fix the `bevy_time` unit tests occasionally failing on optimised Windows
builds.
# Background
I noticed that the `bevy_time` unit tests would fail ~50% of the time
after enabling `opt-level=1` in config.toml, or adding `--release` to
cargo test.
```
> cargo test -p bevy_time --release
thread 'real::test::test_update' panicked at crates\bevy_time\src\real.rs:164:9:
assertion `left != right` failed
left: Some(Instant { t: 9458.0756664s })
right: Some(Instant { t: 9458.0756664s })
```
Disabling optimisations would fix the issue, as would switching from Windows to Linux.
The failing path is roughly:
```rust
let mut time = Time::<Real>::new(Instant::now());
time.update();
time.update();
assert_ne!(time.last_update(), time.first_update());
```
Which kinda boils down to:
```rust
let left = Instant::now();
let right = Instant::now();
assert_ne!(left, right);
```
So the failure seems legit, since there's no guarantee that `Instant::now()` increases between calls.
I suspect it only triggers with a combination of Windows + fast CPU + optimisations (Windows has a lower resolution clock than Linux/MacOS). That would explain why it doesn't fail on the Bevy Github CI (optimisations disabled, and I'm guessing the runner CPUs are clocked lower).
# Solution
Make sure `Instant::now()` has increased before calling `time.update()`.
I also considered:
1. Change the unit tests to accept `Instant:now()` not increasing.
- In retrospect this is maybe the better change?
- There's other unit tests that cover time increasing.
- Could also add a deterministic test for zero delta updates.
- I can switch the PR to this if desired.
2. Avoid any paths that hit `Instant::now()` in unit tests.
- Arguably unit tests should always be deterministic.
- But that would mean a bunch of paths aren't tested.
## Testing
`cargo test -p bevy_time --release`
## System Info
`os: "Windows 10 Pro", kernel: "19045", cpu: "AMD Ryzen 9 7900 12-Core Processor", core_count: "12", memory: "63.2 GiB"`
Also tested on same computer with Linux pop-os 6.9.3.
Co-authored-by: François Mockers <mockersf@gmail.com>
We won't be able to retain render phases from frame to frame if the keys
are unstable. It's not as simple as simply keying off the main world
entity, however, because some main world entities extract to multiple
render world entities. For example, directional lights extract to
multiple shadow cascades, and point lights extract to one view per
cubemap face. Therefore, we key off a new type, `RetainedViewEntity`,
which contains the main entity plus a *subview ID*.
This is part of the preparation for retained bins.
---------
Co-authored-by: ickshonpe <david.curthoys@googlemail.com>
# Objective
I have an application where I'd like to measure average frame rate over
the entire life of the application, and it would be handy if I could
just configure this on the existing `FrameTimeDiagnosticsPlugin`.
Probably fixes#10948?
## Solution
Add `max_history_length` to `FrameTimeDiagnosticsPlugin`, and because
`smoothing_factor` seems to be based on history length, add that too.
## Discussion
I'm not totally sure that `DEFAULT_MAX_HISTORY_LENGTH` is a great
default for `FrameTimeDiagnosticsPlugin` (or any diagnostic?). That's
1/3 of a second at typical game frame rates. Moreover, the default print
interval for `LogDiagnosticsPlugin` is 1 second. So when the two are
combined, you are printing the average over the last third of the
duration between now and the previous print, which seems a bit wonky.
(related: #11429)
I'm pretty sure this default value discussed and the current value
wasn't totally arbitrary though.
Maybe it would be nice for `Diagnostic` to have a
`with_max_history_length_and_also_calculate_a_good_default_smoothing_factor`
method? And then make an explicit smoothing factor in
`FrameTimeDiagnosticsPlugin` optional?
Or add a `new(max_history_length: usize)` method to
`FrameTimeDiagnosticsPlugin` that sets a reasonable default
`smoothing_factor`? edit: This one seems like a no-brainer, doing it.
## Alternatives
It's really easy to roll your own `FrameTimeDiagnosticsPlugin`, but that
might not be super interoperable with, for example, third party FPS
overlays. Still, might be the right call.
## Testing
`cargo run --example many_sprites` (modified to use a custom
`max_history_length`)
## Migration Guide
`FrameTimeDiagnosticsPlugin` now contains two fields. Use
`FrameTimeDiagnosticsPlugin::default()` to match Bevy's previous
behavior or, for example, `FrameTimeDiagnosticsPlugin::new(60)` to
configure it.
