# Objective
- Part of #20115
We want to encapsulate each part of `ScheduleGraph` into its own
specific struct to make parts of it easier to reuse and maintain.
## Solution
- Pulled `ScheduleGraph::systems` and `ScheduleGraph::system_conditions`
into a `Systems` struct and added a field for this new struct to
`ScheduleGraph`
- Broke up `ScheduleGraph::uninit` into `Systems::uninit` and
`SystemSets::uninit` to eliminate `ScheduleGraph`'s direct field access
of these types
- Removed node and condition accessors from `ScheduleGraph`; the same
operations are now available on `Systems` and `SystemSets` instead
(accessible via their `pub` fields on `ScheduleGraph`)
- Moved `Systems`, `SystemSets`, `SystemNode`, `SystemWithAccess`, and
`ConditionWithAccess` into a separate file.
## Testing
Added two new tests covering the API surface of `Systems` and
`SystemSets`, respectively.
---------
Co-authored-by: Chris Russell <8494645+chescock@users.noreply.github.com>
# Objective
> I think we should axe the shared `Event` trait entirely
It doesn't serve any functional purpose, and I don't think it's useful
pedagogically
@alice-i-cecile on discord
## Solution
- Remove `Event` as a supertrait of `BufferedEvent`
- Remove any `Event` derives that were made unnecessary
- Update release notes
---------
Co-authored-by: SpecificProtagonist <vincentjunge@posteo.net>
# Objective
`SystemSetNode` doesn't really add much value beyond a couple helper
functions for getting the name as a string and checking if its a
`SystemTypeSet`. We can replace it entirely and use `InternedSystemSet`
directly by inlining these helper functions' usages without sacrificing
readability.
## Solution
Remove it and replace it with direct `InternedSystemSet` usage.
## Testing
Reusing current tests.
# Objective
Clean up documentation around `UninitializedId`, which has slightly
confusing semantics.
## Solution
Added documentation comments on `UninitializedId`.
---------
Co-authored-by: Chris Russell <8494645+chescock@users.noreply.github.com>
# Objective
Make the schedule graph code more understandable, and replace some
panics with `Result`s.
I found the `check_edges` and `check_hierarchy` functions [a little
confusing](https://github.com/bevyengine/bevy/pull/19352#discussion_r2181486099),
as they combined two concerns: Initializing graph nodes for system sets,
and checking for self-edges on system sets. It was hard to understand
the self-edge checks because it wasn't clear what `NodeId` was being
checked against! So, let's separate those concerns, and move them to
more appropriate locations.
Fix a bug where `schedule.configure_sets((SameSet, SameSet).chain());`
would panic with an unhelpful message: `assertion failed: index_a <
index_b`.
## Solution
Remove the `check_edges` and `check_hierarchy` functions, separating the
initialization behavior and the checking behavior and moving them where
they are easier to understand.
For initializing graph nodes, do this on-demand using the `entry` API by
replacing later `self.system_set_ids[&set]` calls with a
`self.system_sets.get_or_add_set(set)` method. This should avoid the
need for an extra pass over the graph and an extra lookup.
Unfortunately, implementing that method directly on `ScheduleGraph`
leads to borrowing errors as it borrows the entire `struct`. So, split
out the collections managing system sets into a separate `struct`.
For checking self-edges, move this check later so that it can be
reported by returning a `Result` from `Schedule::initialize` instead of
having to panic in `configure_set_inner`. The issue was that `iter_sccs`
does not report self-edges as cycles, since the resulting components
only have one node, but that later code assumes all edges point forward.
So, check for self-edges directly, immediately before calling
`iter_sccs`.
This also ensures we catch *every* way that self-edges can be added. The
previous code missed an edge case where `chain()`ing a set to itself
would create a self-edge and would trigger a `debug_assert`.
# Objective
Allow combinator and pipe systems to delay validation of the second
system, while still allowing the second system to be skipped.
Fixes#18796
Allow fallible systems to be used as one-shot systems, reporting errors
to the error handler when used through commands.
Fixes#19722
Allow fallible systems to be used as run conditions, including when used
with combinators. Alternative to #19580.
Always validate parameters when calling the safe
`run_without_applying_deferred`, `run`, and `run_readonly` methods on a
`System`.
## Solution
Have `System::run_unsafe` return a `Result`.
We want pipe systems to run the first system before validating the
second, since the first system may affect whether the second system has
valid parameters. But if the second system skips then we have no output
value to return! So, pipe systems must return a `Result` that indicates
whether the second system ran.
But if we just make pipe systems have `Out = Result<B::Out>`, then
chaining `a.pipe(b).pipe(c)` becomes difficult. `c` would need to accept
the `Result` from `a.pipe(b)`, which means it would likely need to
return `Result` itself, giving `Result<Result<Out>>`!
Instead, we make *all* systems return a `Result`! We move the handling
of fallible systems from `IntoScheduleConfigs` and `IntoObserverSystem`
to `SystemParamFunction` and `ExclusiveSystemParamFunction`, so that an
infallible system can be wrapped before being passed to a combinator.
As a side effect, this enables fallible systems to be used as run
conditions and one-shot systems.
Now that the safe `run_without_applying_deferred`, `run`, and
`run_readonly` methods return a `Result`, we can have them perform
parameter validation themselves instead of requiring each caller to
remember to call them. `run_unsafe` will continue to not validate
parameters, since it is used in the multi-threaded executor when we want
to validate and run in separate tasks.
Note that this makes type inference a little more brittle. A function
that returns `Result<T>` can be considered either a fallible system
returning `T` or an infallible system returning `Result<T>` (and this is
important to continue supporting `pipe`-based error handling)! So there
are some cases where the output type of a system can no longer be
inferred. It will work fine when directly adding to a schedule, since
then the output type is fixed to `()` (or `bool` for run conditions).
And it will work fine when `pipe`ing to a system with a typed input
parameter.
I used a dedicated `RunSystemError` for the error type instead of plain
`BevyError` so that skipping a system does not box an error or capture a
backtrace.
# Objective
- First step towards #279
## Solution
Makes the necessary internal data structure changes in order to allow
system removal to be added in a future PR: `Vec`s storing systems and
system sets in `ScheduleGraph` have been replaced with `SlotMap`s.
See the included migration guide for the required changes.
## Testing
Internal changes only and no new features *should* mean no new tests are
requried.
# Objective
Fix https://github.com/bevyengine/bevy/issues/19617
# Solution
Add newlines before all impl blocks.
I suspect that at least some of these will be objectionable! If there's
a desired Bevy style for this then I'll update the PR. If not then we
can just close it - it's the work of a single find and replace.
Without this dependency, the bevy_ecs tests fail with missing as_string
methods.
# Objective
- Fixes#19734
## Solution
- add bevy_utils with feature = "Debug" to dev-dependencies
## Testing
- Ran `cargo test -p bevy_ecs`
- Ran `taplo fmt --check`
---
# Objective
- Many strings in bevy_ecs are created but only used for debug: system
name, component name, ...
- Those strings make a significant part of the final binary and are no
use in a released game
## Solution
- Use [`strings`](https://linux.die.net/man/1/strings) to find ...
strings in a binary
- Try to find where they come from
- Many are made from `type_name::<T>()` and only used in error / debug
messages
- Add a new structure `DebugName` that holds no value if `debug` feature
is disabled
- Replace `core::any::type_name::<T>()` by `DebugName::type_name::<T>()`
## Testing
Measurements were taken without the new feature being enabled by
default, to help with commands
### File Size
I tried building the `breakout` example with `cargo run --release
--example breakout`
|`debug` enabled|`debug` disabled|
|-|-|
|81621776 B|77735728B|
|77.84MB|74.13MB|
### Compilation time
`hyperfine --min-runs 15 --prepare "cargo clean && sleep 5"
'RUSTC_WRAPPER="" cargo build --release --example breakout'
'RUSTC_WRAPPER="" cargo build --release --example breakout --features
debug'`
```
breakout' 'RUSTC_WRAPPER="" cargo build --release --example breakout --features debug'
Benchmark 1: RUSTC_WRAPPER="" cargo build --release --example breakout
Time (mean ± σ): 84.856 s ± 3.565 s [User: 1093.817 s, System: 32.547 s]
Range (min … max): 78.038 s … 89.214 s 15 runs
Benchmark 2: RUSTC_WRAPPER="" cargo build --release --example breakout --features debug
Time (mean ± σ): 92.303 s ± 2.466 s [User: 1193.443 s, System: 33.803 s]
Range (min … max): 90.619 s … 99.684 s 15 runs
Summary
RUSTC_WRAPPER="" cargo build --release --example breakout ran
1.09 ± 0.05 times faster than RUSTC_WRAPPER="" cargo build --release --example breakout --features debug
```
# Objective
Closes#19564.
The current `Event` trait looks like this:
```rust
pub trait Event: Send + Sync + 'static {
type Traversal: Traversal<Self>;
const AUTO_PROPAGATE: bool = false;
fn register_component_id(world: &mut World) -> ComponentId { ... }
fn component_id(world: &World) -> Option<ComponentId> { ... }
}
```
The `Event` trait is used by both buffered events
(`EventReader`/`EventWriter`) and observer events. If they are observer
events, they can optionally be targeted at specific `Entity`s or
`ComponentId`s, and can even be propagated to other entities.
However, there has long been a desire to split the trait semantically
for a variety of reasons, see #14843, #14272, and #16031 for discussion.
Some reasons include:
- It's very uncommon to use a single event type as both a buffered event
and targeted observer event. They are used differently and tend to have
distinct semantics.
- A common footgun is using buffered events with observers or event
readers with observer events, as there is no type-level error that
prevents this kind of misuse.
- #19440 made `Trigger::target` return an `Option<Entity>`. This
*seriously* hurts ergonomics for the general case of entity observers,
as you need to `.unwrap()` each time. If we could statically determine
whether the event is expected to have an entity target, this would be
unnecessary.
There's really two main ways that we can categorize events: push vs.
pull (i.e. "observer event" vs. "buffered event") and global vs.
targeted:
| | Push | Pull |
| ------------ | --------------- | --------------------------- |
| **Global** | Global observer | `EventReader`/`EventWriter` |
| **Targeted** | Entity observer | - |
There are many ways to approach this, each with their tradeoffs.
Ultimately, we kind of want to split events both ways:
- A type-level distinction between observer events and buffered events,
to prevent people from using the wrong kind of event in APIs
- A statically designated entity target for observer events to avoid
accidentally using untargeted events for targeted APIs
This PR achieves these goals by splitting event traits into `Event`,
`EntityEvent`, and `BufferedEvent`, with `Event` being the shared trait
implemented by all events.
## `Event`, `EntityEvent`, and `BufferedEvent`
`Event` is now a very simple trait shared by all events.
```rust
pub trait Event: Send + Sync + 'static {
// Required for observer APIs
fn register_component_id(world: &mut World) -> ComponentId { ... }
fn component_id(world: &World) -> Option<ComponentId> { ... }
}
```
You can call `trigger` for *any* event, and use a global observer for
listening to the event.
```rust
#[derive(Event)]
struct Speak {
message: String,
}
// ...
app.add_observer(|trigger: On<Speak>| {
println!("{}", trigger.message);
});
// ...
commands.trigger(Speak {
message: "Y'all like these reworked events?".to_string(),
});
```
To allow an event to be targeted at entities and even propagated
further, you can additionally implement the `EntityEvent` trait:
```rust
pub trait EntityEvent: Event {
type Traversal: Traversal<Self>;
const AUTO_PROPAGATE: bool = false;
}
```
This lets you call `trigger_targets`, and to use targeted observer APIs
like `EntityCommands::observe`:
```rust
#[derive(Event, EntityEvent)]
#[entity_event(traversal = &'static ChildOf, auto_propagate)]
struct Damage {
amount: f32,
}
// ...
let enemy = commands.spawn((Enemy, Health(100.0))).id();
// Spawn some armor as a child of the enemy entity.
// When the armor takes damage, it will bubble the event up to the enemy.
let armor_piece = commands
.spawn((ArmorPiece, Health(25.0), ChildOf(enemy)))
.observe(|trigger: On<Damage>, mut query: Query<&mut Health>| {
// Note: `On::target` only exists because this is an `EntityEvent`.
let mut health = query.get(trigger.target()).unwrap();
health.0 -= trigger.amount();
});
commands.trigger_targets(Damage { amount: 10.0 }, armor_piece);
```
> [!NOTE]
> You *can* still also trigger an `EntityEvent` without targets using
`trigger`. We probably *could* make this an either-or thing, but I'm not
sure that's actually desirable.
To allow an event to be used with the buffered API, you can implement
`BufferedEvent`:
```rust
pub trait BufferedEvent: Event {}
```
The event can then be used with `EventReader`/`EventWriter`:
```rust
#[derive(Event, BufferedEvent)]
struct Message(String);
fn write_hello(mut writer: EventWriter<Message>) {
writer.write(Message("I hope these examples are alright".to_string()));
}
fn read_messages(mut reader: EventReader<Message>) {
// Process all buffered events of type `Message`.
for Message(message) in reader.read() {
println!("{message}");
}
}
```
In summary:
- Need a basic event you can trigger and observe? Derive `Event`!
- Need the event to be targeted at an entity? Derive `EntityEvent`!
- Need the event to be buffered and support the
`EventReader`/`EventWriter` API? Derive `BufferedEvent`!
## Alternatives
I'll now cover some of the alternative approaches I have considered and
briefly explored. I made this section collapsible since it ended up
being quite long :P
<details>
<summary>Expand this to see alternatives</summary>
### 1. Unified `Event` Trait
One option is not to have *three* separate traits (`Event`,
`EntityEvent`, `BufferedEvent`), and to instead just use associated
constants on `Event` to determine whether an event supports targeting
and buffering or not:
```rust
pub trait Event: Send + Sync + 'static {
type Traversal: Traversal<Self>;
const AUTO_PROPAGATE: bool = false;
const TARGETED: bool = false;
const BUFFERED: bool = false;
fn register_component_id(world: &mut World) -> ComponentId { ... }
fn component_id(world: &World) -> Option<ComponentId> { ... }
}
```
Methods can then use bounds like `where E: Event<TARGETED = true>` or
`where E: Event<BUFFERED = true>` to limit APIs to specific kinds of
events.
This would keep everything under one `Event` trait, but I don't think
it's necessarily a good idea. It makes APIs harder to read, and docs
can't easily refer to specific types of events. You can also create
weird invariants: what if you specify `TARGETED = false`, but have
`Traversal` and/or `AUTO_PROPAGATE` enabled?
### 2. `Event` and `Trigger`
Another option is to only split the traits between buffered events and
observer events, since that is the main thing people have been asking
for, and they have the largest API difference.
If we did this, I think we would need to make the terms *clearly*
separate. We can't really use `Event` and `BufferedEvent` as the names,
since it would be strange that `BufferedEvent` doesn't implement
`Event`. Something like `ObserverEvent` and `BufferedEvent` could work,
but it'd be more verbose.
For this approach, I would instead keep `Event` for the current
`EventReader`/`EventWriter` API, and call the observer event a
`Trigger`, since the "trigger" terminology is already used in the
observer context within Bevy (both as a noun and a verb). This is also
what a long [bikeshed on
Discord](https://discord.com/channels/691052431525675048/749335865876021248/1298057661878898791)
seemed to land on at the end of last year.
```rust
// For `EventReader`/`EventWriter`
pub trait Event: Send + Sync + 'static {}
// For observers
pub trait Trigger: Send + Sync + 'static {
type Traversal: Traversal<Self>;
const AUTO_PROPAGATE: bool = false;
const TARGETED: bool = false;
fn register_component_id(world: &mut World) -> ComponentId { ... }
fn component_id(world: &World) -> Option<ComponentId> { ... }
}
```
The problem is that "event" is just a really good term for something
that "happens". Observers are rapidly becoming the more prominent API,
so it'd be weird to give them the `Trigger` name and leave the good
`Event` name for the less common API.
So, even though a split like this seems neat on the surface, I think it
ultimately wouldn't really work. We want to keep the `Event` name for
observer events, and there is no good alternative for the buffered
variant. (`Message` was suggested, but saying stuff like "sends a
collision message" is weird.)
### 3. `GlobalEvent` + `TargetedEvent`
What if instead of focusing on the buffered vs. observed split, we
*only* make a distinction between global and targeted events?
```rust
// A shared event trait to allow global observers to work
pub trait Event: Send + Sync + 'static {
fn register_component_id(world: &mut World) -> ComponentId { ... }
fn component_id(world: &World) -> Option<ComponentId> { ... }
}
// For buffered events and non-targeted observer events
pub trait GlobalEvent: Event {}
// For targeted observer events
pub trait TargetedEvent: Event {
type Traversal: Traversal<Self>;
const AUTO_PROPAGATE: bool = false;
}
```
This is actually the first approach I implemented, and it has the neat
characteristic that you can only use non-targeted APIs like `trigger`
with a `GlobalEvent` and targeted APIs like `trigger_targets` with a
`TargetedEvent`. You have full control over whether the entity should or
should not have a target, as they are fully distinct at the type-level.
However, there's a few problems:
- There is no type-level indication of whether a `GlobalEvent` supports
buffered events or just non-targeted observer events
- An `Event` on its own does literally nothing, it's just a shared trait
required to make global observers accept both non-targeted and targeted
events
- If an event is both a `GlobalEvent` and `TargetedEvent`, global
observers again have ambiguity on whether an event has a target or not,
undermining some of the benefits
- The names are not ideal
### 4. `Event` and `EntityEvent`
We can fix some of the problems of Alternative 3 by accepting that
targeted events can also be used in non-targeted contexts, and simply
having the `Event` and `EntityEvent` traits:
```rust
// For buffered events and non-targeted observer events
pub trait Event: Send + Sync + 'static {
fn register_component_id(world: &mut World) -> ComponentId { ... }
fn component_id(world: &World) -> Option<ComponentId> { ... }
}
// For targeted observer events
pub trait EntityEvent: Event {
type Traversal: Traversal<Self>;
const AUTO_PROPAGATE: bool = false;
}
```
This is essentially identical to this PR, just without a dedicated
`BufferedEvent`. The remaining major "problem" is that there is still
zero type-level indication of whether an `Event` event *actually*
supports the buffered API. This leads us to the solution proposed in
this PR, using `Event`, `EntityEvent`, and `BufferedEvent`.
</details>
## Conclusion
The `Event` + `EntityEvent` + `BufferedEvent` split proposed in this PR
aims to solve all the common problems with Bevy's current event model
while keeping the "weirdness" factor minimal. It splits in terms of both
the push vs. pull *and* global vs. targeted aspects, while maintaining a
shared concept for an "event".
### Why I Like This
- The term "event" remains as a single concept for all the different
kinds of events in Bevy.
- Despite all event types being "events", they use fundamentally
different APIs. Instead of assuming that you can use an event type with
any pattern (when only one is typically supported), you explicitly opt
in to each one with dedicated traits.
- Using separate traits for each type of event helps with documentation
and clearer function signatures.
- I can safely make assumptions on expected usage.
- If I see that an event is an `EntityEvent`, I can assume that I can
use `observe` on it and get targeted events.
- If I see that an event is a `BufferedEvent`, I can assume that I can
use `EventReader` to read events.
- If I see both `EntityEvent` and `BufferedEvent`, I can assume that
both APIs are supported.
In summary: This allows for a unified concept for events, while limiting
the different ways to use them with opt-in traits. No more guess-work
involved when using APIs.
### Problems?
- Because `BufferedEvent` implements `Event` (for more consistent
semantics etc.), you can still use all buffered events for non-targeted
observers. I think this is fine/good. The important part is that if you
see that an event implements `BufferedEvent`, you know that the
`EventReader`/`EventWriter` API should be supported. Whether it *also*
supports other APIs is secondary.
- I currently only support `trigger_targets` for an `EntityEvent`.
However, you can technically target components too, without targeting
any entities. I consider that such a niche and advanced use case that
it's not a huge problem to only support it for `EntityEvent`s, but we
could also split `trigger_targets` into `trigger_entities` and
`trigger_components` if we wanted to (or implement components as
entities :P).
- You can still trigger an `EntityEvent` *without* targets. I consider
this correct, since `Event` implements the non-targeted behavior, and
it'd be weird if implementing another trait *removed* behavior. However,
it does mean that global observers for entity events can technically
return `Entity::PLACEHOLDER` again (since I got rid of the
`Option<Entity>` added in #19440 for ergonomics). I think that's enough
of an edge case that it's not a huge problem, but it is worth keeping in
mind.
- ~~Deriving both `EntityEvent` and `BufferedEvent` for the same type
currently duplicates the `Event` implementation, so you instead need to
manually implement one of them.~~ Changed to always requiring `Event` to
be derived.
## Related Work
There are plans to implement multi-event support for observers,
especially for UI contexts. [Cart's
example](https://github.com/bevyengine/bevy/issues/14649#issuecomment-2960402508)
API looked like this:
```rust
// Truncated for brevity
trigger: Trigger<(
OnAdd<Pressed>,
OnRemove<Pressed>,
OnAdd<InteractionDisabled>,
OnRemove<InteractionDisabled>,
OnInsert<Hovered>,
)>,
```
I believe this shouldn't be in conflict with this PR. If anything, this
PR might *help* achieve the multi-event pattern for entity observers
with fewer footguns: by statically enforcing that all of these events
are `EntityEvent`s in the context of `EntityCommands::observe`, we can
avoid misuse or weird cases where *some* events inside the trigger are
targeted while others are not.
Follow-up of #19274.
Make the `check_change_tick` methods, of which some are now public, take
`CheckChangeTicks` to make it obvious where this tick comes from, see
other PR.
This also affects the `System` trait, hence the many changed files.
---------
Co-authored-by: Chris Russell <8494645+chescock@users.noreply.github.com>
# Objective
Reduce memory usage by storing fewer copies of
`FilteredAccessSet<ComponentId>`.
Currently, the `System` trait exposes the `component_access_set` for the
system, which is used by the multi-threaded executor to determine which
systems can run concurrently. But because it is available on the trait,
it needs to be stored for *every* system, even ones that are not run by
the executor! In particular, it is never needed for observers, or for
the inner systems in a `PipeSystem` or `CombinatorSystem`.
## Solution
Instead of exposing the access from a method on `System`, return it from
`System::initialize`. Since it is still needed during scheduling, store
the access alongside the boxed system in the schedule.
That's not quite enough for systems built using `SystemParamBuilder`s,
though. Those calculate the access in `SystemParamBuilder::build`, which
happens earlier than `System::initialize`. To handle those, we separate
`SystemParam::init_state` into `init_state`, which creates the state
value, and `init_access`, which calculates the access. This lets
`System::initialize` call `init_access` on a state that was provided by
the builder.
An additional benefit of that separation is that it removes the need to
duplicate access checks between `SystemParamBuilder::build` and
`SystemParam::init_state`.
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
# Objective
I set out with one simple goal: clearly document the differences between
each of the component lifecycle events via module docs.
Unfortunately, no such module existed: the various lifecycle code was
scattered to the wind.
Without a unified module, it's very hard to discover the related types,
and there's nowhere good to put my shiny new documentation.
## Solution
1. Unify the assorted types into a single
`bevy_ecs::component_lifecycle` module.
2. Write docs.
3. Write a migration guide.
## Testing
Thanks CI!
## Follow-up
1. The lifecycle event names are pretty confusing, especially
`OnReplace`. We should consider renaming those. No bikeshedding in my PR
though!
2. Observers need real module docs too :(
3. Any additional functional changes should be done elsewhere; this is a
simple docs and re-org PR.
---------
Co-authored-by: theotherphil <phil.j.ellison@gmail.com>
# Objective
Fixes#19403
As described in the issue, the objective is to support the use of
systems returning `Result<(), BevyError>` and
`Result<bool, BevyError>` as run conditions. In these cases, the run
condition would hold on `Ok(())` and `Ok(true)` respectively.
## Solution
`IntoSystem<In, bool, M>` cannot be implemented for systems returning
`Result<(), BevyError>` and `Result<bool, BevyError>` as that would
conflict with their trivial implementation of the trait. That led me to
add a method to the sealed trait `SystemCondition` that does the
conversion. In the original case of a system returning `bool`, the
system is returned as is. With the new types, the system is combined
with `map()` to obtain a `bool`.
By the way, I'm confused as to why `SystemCondition` has a generic `In`
parameter as it is only ever used with `In = ()` as far as I can tell.
## Testing
I added a simple test for both type of system. That's minimal but it
felt enough. I could not picture the more complicated tests passing for
a run condition returning `bool` and failing for the new types.
## Doc
I documenting the change on the page of the trait. I had trouble wording
it right but I'm not sure how to improve it. The phrasing "the condition
returns `true`" which reads naturally is now technically incorrect as
the new types return a `Result`. However, the underlying condition
system that the implementing system turns into does indeed return
`bool`. But talking about the implementation details felt too much.
Another possibility is to use another turn of phrase like "the condition
holds" or "the condition checks out". I've left "the condition returns
`true`" in the documentation of `run_if` and the provided methods for
now.
I'm perplexed about the examples. In the first one, why not implement
the condition directly instead of having a system returning it? Is it
from a time of Bevy where you had to implement your conditions that way?
In that case maybe that should be updated. And in the second example I'm
missing the point entirely. As I stated above, I've only seen conditions
used in contexts where they have no input parameter. Here we create a
condition with an input parameter (cannot be used by `run_if`) and we
are using it with `pipe()` which actually doesn't need our system to
implement `SystemCondition`. Both examples are also calling
`IntoSystem::into_system` which should not be encouraged. What am I
missing?
# Objective
- Cleanup related to #19495.
## Solution
- Delete `System::component_access()`. It is redundant with
`System::component_access_set().combined_access()`.
## Testing
- None. There are no callers of this function.
# Objective
In the past I had custom data structures containing `Tick`s. I learned
that these need to be regularly checked to clamp them. But there was no
way to hook into that logic so I abandoned storing ticks since then.
Another motivation to open this up some more is to be more able to do a
correct implementation of `System::check_ticks`.
## Solution
Add `CheckChangeTicks` and trigger it in `World::check_change_ticks`.
Make `Tick::check_tick` public.
This event makes it possible to store ticks in components or resources
and have them checked.
I also made `Schedules::check_change_ticks` public so users can store
schedules in custom resources/components for whatever reasons.
## Testing
The logic boils down to a single `World::trigger` call and I don't think
this needs more tests.
## Alternatives
Making this obsolete like with #15683.
---
## Showcase
From the added docs:
```rs
use bevy_ecs::prelude::*;
use bevy_ecs::component::CheckChangeTicks;
#[derive(Resource)]
struct CustomSchedule(Schedule);
let mut world = World::new();
world.add_observer(|tick: Trigger<CheckChangeTicks>, mut schedule: ResMut<CustomSchedule>| {
schedule.0.check_change_ticks(tick.get());
});
```
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
# Objective
`SystemSet`s are surprisingly rich and nuanced, but are extremely poorly
documented.
Fixes#19536.
## Solution
Explain the basic concept of system sets, how to create them, and give
some opinionated advice about their more advanced functionality.
## Follow-up
I'd like proper module level docs on system ordering that I can link to
here, but they don't exist. Punting to follow-up!
---------
Co-authored-by: theotherphil <phil.j.ellison@gmail.com>
# Objective
- Fixes#4381
## Solution
- Replace `component_access` with `component_access_set` when
determining conflicting systems during schedule building.
- All `component_access()` impls just forward to
`&component_access_set().combined_access`, so we are essentially trading
`Access::is_compatible` for `FilteredAccessSet::is_compatible`.
- `FilteredAccessSet::get_conflicts` internally calls
`combined_access.is_compatible` as the first step, so we can remove that
redundant check.
## Testing
- Un-ignored a previously failing test now that it passes!
- Ran the `build_schedule` benchmark and got basically no change in the
results. Perhaps are benchmarks are just not targetted towards this
situation.
```
$ critcmp main fix-ambiguity -f 'build_schedule'
group fix-ambiguity main
----- ------------- ----
build_schedule/1000_schedule 1.00 2.9±0.02s ? ?/sec 1.01 2.9±0.05s ? ?/sec
build_schedule/1000_schedule_no_constraints 1.02 48.3±1.48ms ? ?/sec 1.00 47.4±1.78ms ? ?/sec
build_schedule/100_schedule 1.00 9.9±0.17ms ? ?/sec 1.06 10.5±0.32ms ? ?/sec
build_schedule/100_schedule_no_constraints 1.00 804.7±21.85µs ? ?/sec 1.03 828.7±19.36µs ? ?/sec
build_schedule/500_schedule 1.00 451.7±7.25ms ? ?/sec 1.04 468.9±11.70ms ? ?/sec
build_schedule/500_schedule_no_constraints 1.02 12.7±0.46ms ? ?/sec 1.00 12.5±0.44ms ? ?/sec
```
# Objective
- A preparation for the 'system as entities'
- The current system has a series of states such as `is_send`,
`is_exclusive`, `has_defered`, As `system as entites` landed, it may
have more states. Using Bitflags to unify all states is a more concise
and performant approach
## Solution
- Using Bitflags to unify system state.
# Objective
- Enable hot patching systems with subsecond
- Fixes#19296
## Solution
- First commit is the naive thin layer
- Second commit only check the jump table when the code is hot patched
instead of on every system execution
- Depends on https://github.com/DioxusLabs/dioxus/pull/4153 for a nicer
API, but could be done without
- Everything in second commit is feature gated, it has no impact when
the feature is not enabled
## Testing
- Check dependencies without the feature enabled: nothing dioxus in tree
- Run the new example: text and color can be changed
---------
Co-authored-by: Jan Hohenheim <jan@hohenheim.ch>
Co-authored-by: JMS55 <47158642+JMS55@users.noreply.github.com>
# Objective
Fixes https://github.com/bevyengine/bevy/issues/17933
## Solution
Correct "value has changed'" in docs to "value has been added or mutably
dereferenced", with a note for emphasis copied from the docs for
Changed.
## Testing
-
# Objective
There are several uninlined format args (seems to be in more formatting
macros and in more crates) that are not detected on stable, but are on
nightly.
## Solution
Fix them.
# Objective
Remove `ArchetypeComponentId` and `archetype_component_access`.
Following #16885, they are no longer used by the engine, so we can stop
spending time calculating them or space storing them.
## Solution
Remove `ArchetypeComponentId` and everything that touches it.
The `System::update_archetype_component_access` method no longer needs
to update `archetype_component_access`. We do still need to update query
caches, but we no longer need to do so *before* running the system. We'd
have to touch every caller anyway if we gave the method a better name,
so just remove `System::update_archetype_component_access` and
`SystemParam::new_archetype` entirely, and update the query cache in
`Query::get_param`.
The `Single` and `Populated` params also need their query caches updated
in `SystemParam::validate_param`, so change `validate_param` to take
`&mut Self::State` instead of `&Self::State`.
# Objective
- move SyncCell and SyncUnsafeCell to bevy_platform
## Solution
- move SyncCell and SyncUnsafeCell to bevy_platform
## Testing
- cargo clippy works
# Objective
Fixes#18790.
Simpler alternative to #19195.
## Solution
As suggested by @PixelDust22, simply avoid overwriting the pass if the
schedule already has auto sync points enabled.
Leave pass logic untouched.
It still is probably a bad idea to add systems/set configs before
changing the build settings, but that is not important as long there are
no more complex build passes.
## Testing
Added a test.
---------
Co-authored-by: Thierry Berger <contact@thierryberger.com>
Hiya!
# Objective
- Remove upcasting methods that are no longer necessary since Rust 1.86.
- Cleanup the interned label code.
## Notes
- I didn't try to remove the upcasting methods from `bevy_reflect`, as
there appears to be some complexity related to remote type reflection.
- There are likely some other upcasting methods floating around.
## Testing
I ran the `breakout` example to check that the hashing/eq
implementations of the labels are still correct.
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
## Objective
Add documentation useful to users of `bevy_ecs` not also using `App`.
Fixes#19270.
## Solution
* Add explanation of labels to `Schedule` documentation.
* Add example of `derive(ScheduleLabel)` to `trait ScheduleLabel`.
* Add a third example to `Schedule` which demonstrates using a schedule
via label instead of owning it directly.
* Add further explanation and links to `World::add_schedule()`, and
`World::run_schedule()`.
## Testing
Reviewed generated documentation.
Please review this documentation carefully for correctness, as I have
little experience with `bevy_ecs` and I am adding this information
because it would have helped my own past confusion, but I may still be
wrong about how things should be done.
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
Co-authored-by: theotherphil <phil.j.ellison@gmail.com>
# Objective
Fixes#19120
## Solution
Use the find and replace token feature in VSCode to replace all the
`Condition`s with `SystemCondition`s. Then look through all the
documentation with find and replace to replace all the `Condition`s
there.
## Testing
- Did you test these changes? If so, how?
Yes, used cargo clippy, cargo build and cargo test.
- Are there any parts that need more testing?
Nope
- How can other people (reviewers) test your changes? Is there anything
specific they need to know?
By compiling and running bevy
- If relevant, what platforms did you test these changes on, and are
there any important ones you can't test?
Shouldn't be, but Fedora Linux with KDE Wayland
# Objective
[see original
comment](https://github.com/bevyengine/bevy/pull/18801#issuecomment-2796981745)
> Alternately, could we store it on the World instead of a global? I
think we have a World nearby whenever we call default_error_handler().
That would avoid the need for atomics or locks, since we could do
ordinary reads and writes to the World.
Global error handlers don't actually need to be global – per world is
enough. This allows using different handlers for different worlds and
also removes the restrictions on changing the handler only once.
## Solution
Each `World` can now store its own error handler in a resource.
For convenience, you can also set the default error handler for an
`App`, which applies it to the worlds of all `SubApp`s. The old behavior
of only being able to set the error handler once is kept for apps.
We also don't need the `configurable_error_handler` feature anymore now.
## Testing
New/adjusted tests for failing schedule systems & observers.
---
## Showcase
```rust
App::new()
.set_error_handler(info)
…
```
# Objective
A fair few items were deprecated in 0.16. Let's delete them now that
we're in the 0.17 development cycle!
## Solution
- Deleted items marked deprecated in 0.16.
## Testing
- CI
---
## Notes
I'm making the assumption that _everything_ deprecated in 0.16 should be
removed in 0.17. That may be a false assumption in certain cases. Please
check the items to be removed to see if there are any exceptions we
should keep around for another cycle!
# Objective
Fixes a part of #14274.
Bevy has an incredibly inconsistent naming convention for its system
sets, both internally and across the ecosystem.
<img alt="System sets in Bevy"
src="https://github.com/user-attachments/assets/d16e2027-793f-4ba4-9cc9-e780b14a5a1b"
width="450" />
*Names of public system set types in Bevy*
Most Bevy types use a naming of `FooSystem` or just `Foo`, but there are
also a few `FooSystems` and `FooSet` types. In ecosystem crates on the
other hand, `FooSet` is perhaps the most commonly used name in general.
Conventions being so wildly inconsistent can make it harder for users to
pick names for their own types, to search for system sets on docs.rs, or
to even discern which types *are* system sets.
To reign in the inconsistency a bit and help unify the ecosystem, it
would be good to establish a common recommended naming convention for
system sets in Bevy itself, similar to how plugins are commonly suffixed
with `Plugin` (ex: `TimePlugin`). By adopting a consistent naming
convention in first-party Bevy, we can softly nudge ecosystem crates to
follow suit (for types where it makes sense to do so).
Choosing a naming convention is also relevant now, as the [`bevy_cli`
recently adopted
lints](https://github.com/TheBevyFlock/bevy_cli/pull/345) to enforce
naming for plugins and system sets, and the recommended naming used for
system sets is still a bit open.
## Which Name To Use?
Now the contentious part: what naming convention should we actually
adopt?
This was discussed on the Bevy Discord at the end of last year, starting
[here](<https://discord.com/channels/691052431525675048/692572690833473578/1310659954683936789>).
`FooSet` and `FooSystems` were the clear favorites, with `FooSet` very
narrowly winning an unofficial poll. However, it seems to me like the
consensus was broadly moving towards `FooSystems` at the end and after
the poll, with Cart
([source](https://discord.com/channels/691052431525675048/692572690833473578/1311140204974706708))
and later Alice
([source](https://discord.com/channels/691052431525675048/692572690833473578/1311092530732859533))
and also me being in favor of it.
Let's do a quick pros and cons list! Of course these are just what I
thought of, so take it with a grain of salt.
`FooSet`:
- Pro: Nice and short!
- Pro: Used by many ecosystem crates.
- Pro: The `Set` suffix comes directly from the trait name `SystemSet`.
- Pro: Pairs nicely with existing APIs like `in_set` and
`configure_sets`.
- Con: `Set` by itself doesn't actually indicate that it's related to
systems *at all*, apart from the implemented trait. A set of what?
- Con: Is `FooSet` a set of `Foo`s or a system set related to `Foo`? Ex:
`ContactSet`, `MeshSet`, `EnemySet`...
`FooSystems`:
- Pro: Very clearly indicates that the type represents a collection of
systems. The actual core concept, system(s), is in the name.
- Pro: Parallels nicely with `FooPlugins` for plugin groups.
- Pro: Low risk of conflicts with other names or misunderstandings about
what the type is.
- Pro: In most cases, reads *very* nicely and clearly. Ex:
`PhysicsSystems` and `AnimationSystems` as opposed to `PhysicsSet` and
`AnimationSet`.
- Pro: Easy to search for on docs.rs.
- Con: Usually results in longer names.
- Con: Not yet as widely used.
Really the big problem with `FooSet` is that it doesn't actually
describe what it is. It describes what *kind of thing* it is (a set of
something), but not *what it is a set of*, unless you know the type or
check its docs or implemented traits. `FooSystems` on the other hand is
much more self-descriptive in this regard, at the cost of being a bit
longer to type.
Ultimately, in some ways it comes down to preference and how you think
of system sets. Personally, I was originally in favor of `FooSet`, but
have been increasingly on the side of `FooSystems`, especially after
seeing what the new names would actually look like in Avian and now
Bevy. I prefer it because it usually reads better, is much more clearly
related to groups of systems than `FooSet`, and overall *feels* more
correct and natural to me in the long term.
For these reasons, and because Alice and Cart also seemed to share a
preference for it when it was previously being discussed, I propose that
we adopt a `FooSystems` naming convention where applicable.
## Solution
Rename Bevy's system set types to use a consistent `FooSet` naming where
applicable.
- `AccessibilitySystem` → `AccessibilitySystems`
- `GizmoRenderSystem` → `GizmoRenderSystems`
- `PickSet` → `PickingSystems`
- `RunFixedMainLoopSystem` → `RunFixedMainLoopSystems`
- `TransformSystem` → `TransformSystems`
- `RemoteSet` → `RemoteSystems`
- `RenderSet` → `RenderSystems`
- `SpriteSystem` → `SpriteSystems`
- `StateTransitionSteps` → `StateTransitionSystems`
- `RenderUiSystem` → `RenderUiSystems`
- `UiSystem` → `UiSystems`
- `Animation` → `AnimationSystems`
- `AssetEvents` → `AssetEventSystems`
- `TrackAssets` → `AssetTrackingSystems`
- `UpdateGizmoMeshes` → `GizmoMeshSystems`
- `InputSystem` → `InputSystems`
- `InputFocusSet` → `InputFocusSystems`
- `ExtractMaterialsSet` → `MaterialExtractionSystems`
- `ExtractMeshesSet` → `MeshExtractionSystems`
- `RumbleSystem` → `RumbleSystems`
- `CameraUpdateSystem` → `CameraUpdateSystems`
- `ExtractAssetsSet` → `AssetExtractionSystems`
- `Update2dText` → `Text2dUpdateSystems`
- `TimeSystem` → `TimeSystems`
- `AudioPlaySet` → `AudioPlaybackSystems`
- `SendEvents` → `EventSenderSystems`
- `EventUpdates` → `EventUpdateSystems`
A lot of the names got slightly longer, but they are also a lot more
consistent, and in my opinion the majority of them read much better. For
a few of the names I took the liberty of rewording things a bit;
definitely open to any further naming improvements.
There are still also cases where the `FooSystems` naming doesn't really
make sense, and those I left alone. This primarily includes system sets
like `Interned<dyn SystemSet>`, `EnterSchedules<S>`, `ExitSchedules<S>`,
or `TransitionSchedules<S>`, where the type has some special purpose and
semantics.
## Todo
- [x] Should I keep all the old names as deprecated type aliases? I can
do this, but to avoid wasting work I'd prefer to first reach consensus
on whether these renames are even desired.
- [x] Migration guide
- [x] Release notes
# Objective
Contributes to #18741 and #18453.
## Solution
Deprecate `SimpleExecutor`. If users run into migration issues, we can
backtrack. Otherwise, we follow this up with #18741
We can't easily deprecate the module too because of
[this](https://github.com/rust-lang/rust/issues/47238).
## Testing
CI
---------
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
Co-authored-by: Cyrill Schenkel <cyrill.schenkel@gmail.com>
# Objective
Simplify code in the `SingleThreadedExecutor` by removing a special case
for exclusive systems.
The `SingleThreadedExecutor` runs systems without immediately applying
deferred buffers. That required calling `run_unsafe()` instead of
`run()`, but that would `panic` for exclusive systems, so the code also
needed a special case for those. Following #18076 and #18406, we have a
`run_without_applying_deferred` method that has the exact behavior we
want and works on exclusive systems.
## Solution
Replace the code in `SingleThreadedExecutor` that runs systems with a
single call to `run_without_applying_deferred()`. Also add this as a
wrapper in the `__rust_begin_short_backtrace` module to preserve the
special behavior for backtraces.
# Objective
Stop using `ArchetypeComponentId` in the executor. These IDs will grow
even more quickly with relations, and the size may start to degrade
performance.
## Solution
Have systems expose their `FilteredAccessSet<ComponentId>`, and have the
executor use that to determine which systems conflict. This can be
determined statically, so determine all conflicts during initialization
and only perform bit tests when running.
## Testing
I ran many_foxes and didn't see any performance changes. It's probably
worth testing this with a wider range of realistic schedules to see
whether the reduced concurrency has a cost in practice, but I don't know
what sort of test cases to use.
## Migration Guide
The schedule will now prevent systems from running in parallel if there
*could* be an archetype that they conflict on, even if there aren't
actually any. For example, these systems will now conflict even if no
entity has both `Player` and `Enemy` components:
```rust
fn player_system(query: Query<(&mut Transform, &Player)>) {}
fn enemy_system(query: Query<(&mut Transform, &Enemy)>) {}
```
To allow them to run in parallel, use `Without` filters, just as you
would to allow both queries in a single system:
```rust
// Either one of these changes alone would be enough
fn player_system(query: Query<(&mut Transform, &Player), Without<Enemy>>) {}
fn enemy_system(query: Query<(&mut Transform, &Enemy), Without<Player>>) {}
```
# Objective
After #17967, closures which always panic no longer satisfy various Bevy
traits. Principally, this affects observers, systems and commands.
While this may seem pointless (systems which always panic are kind of
useless), it is distinctly annoying when using the `todo!` macro, or
when writing tests that should panic.
Fixes#18778.
## Solution
- Add failing tests to demonstrate the problem
- Add the trick from
[`never_say_never`](https://docs.rs/never-say-never/latest/never_say_never/)
to name the `!` type on stable Rust
- Write looots of docs explaining what the heck is going on and why
we've done this terrible thing
## To do
Unfortunately I couldn't figure out how to avoid conflicting impls, and
I am out of time for today, the week and uh the week after that.
Vacation! If you feel like finishing this for me, please submit PRs to
my branch and I can review and press the button for it while I'm off.
Unless you're Cart, in which case you have write permissions to my
branch!
- [ ] fix for commands
- [ ] fix for systems
- [ ] fix for observers
- [ ] revert https://github.com/bevyengine/bevy-website/pull/2092/
## Testing
I've added a compile test for these failure cases and a few adjacent
non-failing cases (with explicit return types).
---------
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
The goal of `bevy_platform_support` is to provide a set of platform
agnostic APIs, alongside platform-specific functionality. This is a high
traffic crate (providing things like HashMap and Instant). Especially in
light of https://github.com/bevyengine/bevy/discussions/18799, it
deserves a friendlier / shorter name.
Given that it hasn't had a full release yet, getting this change in
before Bevy 0.16 makes sense.
## Solution
- Rename `bevy_platform_support` to `bevy_platform`.
# Objective
- Piped systems are an edge case that we missed when reworking system
parameter validation.
- Fixes#18755.
## Solution
- Validate the parameters for both systems, ~~combining the errors if
both failed validation~~ by simply using an early out.
- ~~Also fix the same bug for combinator systems while we're here.~~
## Testing
I've added a large number of tests checking the behavior under various
permutations. These are separate tests, rather than one mega test
because a) it's easier to track down bugs that way and b) many of these
are `should_panic` tests, which will halt the evaluation of the rest of
the test!
I've also added a test for exclusive systems being pipeable because we
don't have one and I was very surprised that that works!
---------
Co-authored-by: Chris Russell <8494645+chescock@users.noreply.github.com>
# Objective
#18173 allows components to be queued without being fully registered.
But much of bevy's debug logging contained
`components.get_name(id).unwrap()`. However, this panics when the id is
queued. This PR fixes this, allowing names to be retrieved for debugging
purposes, etc, even while they're still queued.
## Solution
We change `ComponentInfo::descriptor` to be `Arc<ComponentDescriptor>`
instead of not arc'd. This lets us pass the descriptor around (as a name
or otherwise) as needed. The alternative would require some form of
`MappedRwLockReadGuard`, which is unstable, and would be terribly
blocking. Putting it in an arc also signifies that it doesn't change,
which is a nice signal to users. This does mean there's an extra pointer
dereference, but I don't think that's an issue here, as almost all paths
that use this are for debugging purposes or one-time set ups.
## Testing
Existing tests.
## Migration Guide
`Components::get_name` now returns `Option<Cow<'_, str>` instead of
`Option<&str>`. This is because it now returns results for queued
components. If that behavior is not desired, or you know the component
is not queued, you can use
`components.get_info().map(ComponentInfo::name)` instead.
Similarly, `ScheduleGraph::conflicts_to_string` now returns `impl
Iterator<Item = (String, String, Vec<Cow<str>>)>` instead of `impl
Iterator<Item = (String, String, Vec<&str>)>`. Because `Cow<str>` derefs
to `&str`, most use cases can remain unchanged.
---------
Co-authored-by: Chris Russell <8494645+chescock@users.noreply.github.com>
# Objective
Fix panic in `run_system` when running an exclusive system wrapped in a
`PipeSystem` or `AdapterSystem`.
#18076 introduced a `System::run_without_applying_deferred` method. It
normally calls `System::run_unsafe`, but
`ExclusiveFunctionSystem::run_unsafe` panics, so it was overridden for
that type. Unfortunately, `PipeSystem::run_without_applying_deferred`
still calls `PipeSystem::run_unsafe`, which can then call
`ExclusiveFunctionSystem::run_unsafe` and panic.
## Solution
Make `ExclusiveFunctionSystem::run_unsafe` work instead of panicking.
Clarify the safety requirements that make this sound.
The alternative is to override `run_without_applying_deferred` in
`PipeSystem`, `CombinatorSystem`, `AdapterSystem`,
`InfallibleSystemWrapper`, and `InfallibleObserverWrapper`. That seems
like a lot of extra code just to preserve a confusing special case!
Remove some implementations of `System::run` that are no longer
necessary with this change. This slightly changes the behavior of
`PipeSystem` and `CombinatorSystem`: Currently `run` will call
`apply_deferred` on the first system before running the second, but
after this change it will only call it after *both* systems have run.
The new behavior is consistent with `run_unsafe` and
`run_without_applying_deferred`, and restores the behavior prior to
#11823.
The panic was originally necessary because [`run_unsafe` took
`&World`](https://github.com/bevyengine/bevy/pull/6083/files#diff-708dfc60ec5eef432b20a6f471357a7ea9bfb254dc2f918d5ed4a66deb0e85baR90).
Now that it takes `UnsafeWorldCell`, it is possible to make it work. See
also Cart's concerns at
https://github.com/bevyengine/bevy/pull/4166#discussion_r979140356,
although those also predate `UnsafeWorldCell`.
And see #6698 for a previous bug caused by this panic.
# Objective
Make it easier to short-circuit system parameter validation.
Simplify the API surface by combining `ValidationOutcome` with
`SystemParamValidationError`.
## Solution
Replace `ValidationOutcome` with `Result<(),
SystemParamValidationError>`. Move the docs from `ValidationOutcome` to
`SystemParamValidationError`.
Add a `skipped` field to `SystemParamValidationError` to distinguish the
`Skipped` and `Invalid` variants.
Use the `?` operator to short-circuit validation in tuples of system
params.
# Objective
When introduced, `Single` was intended to simply be silently skipped,
allowing for graceful and efficient handling of systems during invalid
game states (such as when the player is dead).
However, this also caused missing resources to *also* be silently
skipped, leading to confusing and very hard to debug failures. In
0.15.1, this behavior was reverted to a panic, making missing resources
easier to debug, but largely making `Single` (and `Populated`)
worthless, as they would panic during expected game states.
Ultimately, the consensus is that this behavior should differ on a
per-system-param basis. However, there was no sensible way to *do* that
before this PR.
## Solution
Swap `SystemParam::validate_param` from a `bool` to:
```rust
/// The outcome of system / system param validation,
/// used by system executors to determine what to do with a system.
pub enum ValidationOutcome {
/// All system parameters were validated successfully and the system can be run.
Valid,
/// At least one system parameter failed validation, and an error must be handled.
/// By default, this will result in1 a panic. See [crate::error] for more information.
///
/// This is the default behavior, and is suitable for system params that should *always* be valid,
/// either because sensible fallback behavior exists (like [`Query`] or because
/// failures in validation should be considered a bug in the user's logic that must be immediately addressed (like [`Res`]).
Invalid,
/// At least one system parameter failed validation, but the system should be skipped due to [`ValidationBehavior::Skip`].
/// This is suitable for system params that are intended to only operate in certain application states, such as [`Single`].
Skipped,
}
```
Then, inside of each `SystemParam` implementation, return either Valid,
Invalid or Skipped.
Currently, only `Single`, `Option<Single>` and `Populated` use the
`Skipped` behavior. Other params (like resources) retain their current
failing
## Testing
Messed around with the fallible_params example. Added a pair of tests:
one for panicking when resources are missing, and another for properly
skipping `Single` and `Populated` system params.
## To do
- [x] get https://github.com/bevyengine/bevy/pull/18454 merged
- [x] fix the todo!() in the macro-powered tuple implementation (please
help 🥺)
- [x] test
- [x] write a migration guide
- [x] update the example comments
## Migration Guide
Various system and system parameter validation methods
(`SystemParam::validate_param`, `System::validate_param` and
`System::validate_param_unsafe`) now return and accept a
`ValidationOutcome` enum, rather than a `bool`. The previous `true`
values map to `ValidationOutcome::Valid`, while `false` maps to
`ValidationOutcome::Invalid`.
However, if you wrote a custom schedule executor, you should now respect
the new `ValidationOutcome::Skipped` parameter, skipping any systems
whose validation was skipped. By contrast, `ValidationOutcome::Invalid`
systems should also be skipped, but you should call the
`default_error_handler` on them first, which by default will result in a
panic.
If you are implementing a custom `SystemParam`, you should consider
whether failing system param validation is an error or an expected
state, and choose between `Invalid` and `Skipped` accordingly. In Bevy
itself, `Single` and `Populated` now once again skip the system when
their conditions are not met. This is the 0.15.0 behavior, but stands in
contrast to the 0.15.1 behavior, where they would panic.
---------
Co-authored-by: MiniaczQ <xnetroidpl@gmail.com>
Co-authored-by: Dmytro Banin <banind@cs.washington.edu>
Co-authored-by: Chris Russell <8494645+chescock@users.noreply.github.com>
# Objective
There are two related problems here:
1. Users should be able to change the fallback behavior of *all*
ECS-based errors in their application by setting the
`GLOBAL_ERROR_HANDLER`. See #18351 for earlier work in this vein.
2. The existing solution (#15500) for customizing this behavior is high
on boilerplate, not global and adds a great deal of complexity.
The consensus is that the default behavior when a parameter fails
validation should be set based on the kind of system parameter in
question: `Single` / `Populated` should silently skip the system, but
`Res` should panic. Setting this behavior at the system level is a
bandaid that makes getting to that ideal behavior more painful, and can
mask real failures (if a resource is missing but you've ignored a system
to make the Single stop panicking you're going to have a bad day).
## Solution
I've removed the existing `ParamWarnPolicy`-based configuration, and
wired up the `GLOBAL_ERROR_HANDLER`/`default_error_handler` to the
various schedule executors to properly plumb through errors .
Additionally, I've done a small cleanup pass on the corresponding
example.
## Testing
I've run the `fallible_params` example, with both the default and a
custom global error handler. The former panics (as expected), and the
latter spams the error console with warnings 🥲
## Questions for reviewers
1. Currently, failed system param validation will result in endless
console spam. Do you want me to implement a solution for warn_once-style
debouncing somehow?
2. Currently, the error reporting for failed system param validation is
very limited: all we get is that a system param failed validation and
the name of the system. Do you want me to implement improved error
reporting by bubbling up errors in this PR?
3. There is broad consensus that the default behavior for failed system
param validation should be set on a per-system param basis. Would you
like me to implement that in this PR?
My gut instinct is that we absolutely want to solve 2 and 3, but it will
be much easier to do that work (and review it) if we split the PRs
apart.
## Migration Guide
`ParamWarnPolicy` and the `WithParamWarnPolicy` have been removed
completely. Failures during system param validation are now handled via
the `GLOBAL_ERROR_HANDLER`: please see the `bevy_ecs::error` module docs
for more information.
---------
Co-authored-by: MiniaczQ <xnetroidpl@gmail.com>
# Objective
- ECS error handling is a lovely flagship feature for Bevy 0.16, all in
the name of reducing panics and encouraging better error handling
(#14275).
- Currently though, command and system error handling are completely
disjoint and use different mechanisms.
- Additionally, there's a number of distinct ways to set the
default/fallback/global error handler that have limited value. As far as
I can tell, this will be cfg flagged to toggle between dev and
production builds in 99.9% of cases, with no real value in more granular
settings or helpers.
- Fixes#17272
## Solution
- Standardize error handling on the OnceLock global error mechanisms
ironed out in https://github.com/bevyengine/bevy/pull/17215
- As discussed there, there are serious performance concerns there,
especially for commands
- I also think this is a better fit for the use cases, as it's truly
global
- Move from `SystemErrorContext` to a more general purpose
`ErrorContext`, which can handle observers and commands more clearly
- Cut the superfluous setter methods on `App` and `SubApp`
- Rename the limited (and unhelpful) `fallible_systems` example to
`error_handling`, and add an example of command error handling
## Testing
Ran the `error_handling` example.
## Notes for reviewers
- Do you see a clear way to allow commands to retain &mut World access
in the per-command custom error handlers? IMO that's a key feature here
(allowing the ad-hoc creation of custom commands), but I'm not sure how
to get there without exploding complexity.
- I've removed the feature gate on the default_error_handler: contrary
to @cart's opinion in #17215 I think that virtually all apps will want
to use this. Can you think of a category of app that a) is extremely
performance sensitive b) is fine with shipping to production with the
panic error handler? If so, I can try to gather performance numbers
and/or reintroduce the feature flag. UPDATE: see benches at the end of
this message.
- ~~`OnceLock` is in `std`: @bushrat011899 what should we do here?~~
- Do you have ideas for more automated tests for this collection of
features?
## Benchmarks
I checked the impact of the feature flag introduced: benchmarks might
show regressions. This bears more investigation. I'm still skeptical
that there are users who are well-served by a fast always panicking
approach, but I'm going to re-add the feature flag here to avoid
stalling this out.
---------
Co-authored-by: Zachary Harrold <zac@harrold.com.au>
# Objective
Prevents duplicate implementation between IntoSystemConfigs and
IntoSystemSetConfigs using a generic, adds a NodeType trait for more
config flexibility (opening the door to implement
https://github.com/bevyengine/bevy/issues/14195?).
## Solution
Followed writeup by @ItsDoot:
https://hackmd.io/@doot/rJeefFHc1x
Removes IntoSystemConfigs and IntoSystemSetConfigs, instead using
IntoNodeConfigs with generics.
## Testing
Pending
---
## Showcase
N/A
## Migration Guide
SystemSetConfigs -> NodeConfigs<InternedSystemSet>
SystemConfigs -> NodeConfigs<ScheduleSystem>
IntoSystemSetConfigs -> IntoNodeConfigs<InternedSystemSet, M>
IntoSystemConfigs -> IntoNodeConfigs<ScheduleSystem, M>
---------
Co-authored-by: Christian Hughes <9044780+ItsDoot@users.noreply.github.com>
Co-authored-by: Alice Cecile <alice.i.cecile@gmail.com>
# Objective
- Prevent usage of `println!`, `eprintln!` and the like because they
require `std`
- Fixes#17446
## Solution
- Enable the `print_stdout` and `print_stderr` clippy lints
- Replace all `println!` and `eprintln!` occurrences with `log::*` where
applicable or alternatively ignore the warnings
## Testing
- Run `cargo clippy --workspace` to ensure that there are no warnings
relating to printing to `stdout` or `stderr`
# Objective
This is an alternative to #17871 and #17701 for tracking issue #18155.
This thanks to @maniwani for help with this design.
The goal is to enable component ids to be reserved from multiple threads
concurrently and with only `&World`. This contributes to assets as
entities, read-only query and system parameter initialization, etc.
## What's wrong with #17871 ?
In #17871, I used my proposed staging utilities to allow *fully*
registering components from any thread concurrently with only
`&Components`. However, if we want to pursue components as entities
(which is desirable for a great many reasons. See
[here](https://discord.com/channels/691052431525675048/692572690833473578/1346499196655505534)
on discord), this staging isn't going to work. After all, if registering
a component requires spawning an entity, and spawning an entity requires
`&mut World`, it is impossible to register a component fully with only
`&World`.
## Solution
But what if we don't have to register it all the way? What if it's
enough to just know the `ComponentId` it will have once it is registered
and to queue it to be registered at a later time? Spoiler alert: That is
all we need for these features.
Here's the basic design:
Queue a registration:
1. Check if it has already been registered.
2. Check if it has already been queued.
3. Reserve a `ComponentId`.
4. Queue the registration at that id.
Direct (normal) registration:
1. Check if this registration has been queued.
2. If it has, use the queued registration instead.
3. Otherwise, proceed like normal.
Appllying the queue:
1. Pop queued items off one by one.
2. Register them directly.
One other change:
The whole point of this design over #17871 is to facilitate coupling
component registration with the World. To ensure that this would fully
work with that, I went ahead and moved the `ComponentId` generator onto
the world itself. That stemmed a couple of minor organizational changes
(see migration guide). As we do components as entities, we will replace
this generator with `Entities`, which lives on `World` too. Doing this
move early let me verify the design and will reduce migration headaches
in the future. If components as entities is as close as I think it is, I
don't think splitting this up into different PRs is worth it. If it is
not as close as it is, it might make sense to still do #17871 in the
meantime (see the risks section). I'll leave it up to y'all what we end
up doing though.
## Risks and Testing
The biggest downside of this compared to #17871 is that now we have to
deal with correct but invalid `ComponentId`s. They are invalid because
the component still isn't registered, but they are correct because, once
registered, the component will have exactly that id.
However, the only time this becomes a problem is if some code violates
safety rules by queuing a registration and using the returned id as if
it was valid. As this is a new feature though, nothing in Bevy does
this, so no new tests were added for it. When we do use it, I left
detailed docs to help mitigate issues here, and we can test those
usages. Ex: we will want some tests on using queries initialized from
queued registrations.
## Migration Guide
Component registration can now be queued with only `&World`. To
facilitate this, a few APIs needed to be moved around.
The following functions have moved from `Components` to
`ComponentsRegistrator`:
- `register_component`
- `register_component_with_descriptor`
- `register_resource_with_descriptor`
- `register_non_send`
- `register_resource`
- `register_required_components_manual`
Accordingly, functions in `Bundle` and `Component` now take
`ComponentsRegistrator` instead of `Components`.
You can obtain `ComponentsRegistrator` from the new
`World::components_registrator`.
You can obtain `ComponentsQueuedRegistrator` from the new
`World::components_queue`, and use it to stage component registration if
desired.
# Open Question
Can we verify that it is enough to queue registration with `&World`? I
don't think it would be too difficult to package this up into a
`Arc<MyComponentsManager>` type thing if we need to, but keeping this on
`&World` certainly simplifies things. If we do need the `Arc`, we'll
need to look into partitioning `Entities` for components as entities, so
we can keep most of the allocation fast on `World` and only keep a
smaller partition in the `Arc`. I'd love an SME on assets as entities to
shed some light on this.
---------
Co-authored-by: andriyDev <andriydzikh@gmail.com>
