# Objective

Fixes #8965.

#### Background

For convenience and to ensure everything is setup properly, we
automatically add certain bounds to the derived types. The current
implementation does this by taking the types from all active fields and
adding them to the where-clause of the generated impls. I believe this
method was chosen because it won't add bounds to types that are
otherwise ignored.

```rust
#[derive(Reflect)]
struct Foo<T, U: SomeTrait, V> {
  t: T,
  u: U::Assoc,
  #[reflect(ignore)]
  v: [V; 2]
}

// Generates something like:
impl<T, U: SomeTrait, V> for Foo<T, U, V>
where
  // Active:
  T: Reflect,
  U::Assoc: Reflect,

  // Ignored:
  [V; 2]: Send + Sync + Any
{
  // ...
}
```

The self-referential type fails because it ends up using _itself_ as a
type bound due to being one of its own active fields.

```rust
#[derive(Reflect)]
struct Foo {
  foo: Vec<Foo>
}

// Foo where Vec<Foo>: Reflect -> Vec<T> where T: Reflect -> Foo where Vec<Foo>: Reflect -> ...
```

## Solution

We can't simply parse all field types for the name of our type. That
would be both complex and prone to errors and false-positives. And even
if it wasn't, what would we replace the bound with?

Instead, I opted to go for a solution that only adds the bounds to what
really needs it: the type parameters. While the bounds on concrete types
make errors a bit cleaner, they aren't strictly necessary. This means we
can change our generated where-clause to only add bounds to generic type
parameters.

Doing this, though, returns us back to the problem of over-bounding
parameters that don't need to be bounded. To solve this, I added a new
container attribute (based on
[this](dtolnay/syn#422 (comment))
comment and @nicopap's
[comment](#9046 (comment)))
that allows us to pass in a custom where clause to modify what bounds
are added to these type parameters.

This allows us to do stuff like:

```rust
trait Trait {
  type Assoc;
}

// We don't need `T` to be reflectable since we only care about `T::Assoc`.
#[derive(Reflect)]
#[reflect(where T::Assoc: FromReflect)]
struct Foo<T: Trait>(T::Assoc);

#[derive(TypePath)]
struct Bar;

impl Trait for Bar {
  type Assoc = usize;
}

#[derive(Reflect)]
struct Baz {
  a: Foo<Bar>,
}
```

> **Note**
> I also
[tried](dc139ea)
allowing `#[reflect(ignore)]` to be used on the type parameters
themselves, but that proved problematic since the derive macro does not
consume the attribute. This is why I went with the container attribute
approach.

### Alternatives

One alternative could possibly be to just not add reflection bounds
automatically (i.e. only add required bounds like `Send`, `Sync`, `Any`,
and `TypePath`).

The downside here is we add more friction to using reflection, which
already comes with its own set of considerations. This is a potentially
viable option, but we really need to consider whether or not the
ergonomics hit is worth it.

If we did decide to go the more manual route, we should at least
consider something like #5772 to make it easier for users to add the
right bounds (although, this could still become tricky with
`FromReflect` also being automatically derived).

### Open Questions

1. Should we go with this approach or the manual alternative?
2. ~~Should we add a `skip_params` attribute to avoid the `T: 'static`
trick?~~ ~~Decided to go with `custom_where()` as it's the simplest~~
Scratch that, went with a normal where clause
3. ~~`custom_where` bikeshedding?~~ No longer needed since we are using
a normal where clause

### TODO

- [x] Add compile-fail tests

---

## Changelog

- Fixed issue preventing recursive types from deriving `Reflect`
- Changed how where-clause bounds are generated by the `Reflect` derive
macro
- They are now only applied to the type parameters, not to all active
fields
- Added `#[reflect(where T: Trait, U::Assoc: Trait, ...)]` container
attribute

## Migration Guide

When deriving `Reflect`, generic type params that do not need the
automatic reflection bounds (such as `Reflect`) applied to them will
need to opt-out using a custom where clause like: `#[reflect(where T:
Trait, U::Assoc: Trait, ...)]`.

The attribute can define custom bounds only used by the reflection
impls. To simply opt-out all the type params, we can pass in an empty
where clause: `#[reflect(where)]`.

```rust
// BEFORE:
#[derive(Reflect)]
struct Foo<T>(#[reflect(ignore)] T);

// AFTER:
#[derive(Reflect)]
#[reflect(where)]
struct Foo<T>(#[reflect(ignore)] T);
```

---------

Co-authored-by: Nicola Papale <[email protected]>

# Objective

Fixes bevyengine#8965.

#### Background

For convenience and to ensure everything is setup properly, we
automatically add certain bounds to the derived types. The current
implementation does this by taking the types from all active fields and
adding them to the where-clause of the generated impls. I believe this
method was chosen because it won't add bounds to types that are
otherwise ignored.

```rust
#[derive(Reflect)]
struct Foo<T, U: SomeTrait, V> {
  t: T,
  u: U::Assoc,
  #[reflect(ignore)]
  v: [V; 2]
}

// Generates something like:
impl<T, U: SomeTrait, V> for Foo<T, U, V>
where
  // Active:
  T: Reflect,
  U::Assoc: Reflect,

  // Ignored:
  [V; 2]: Send + Sync + Any
{
  // ...
}
```

The self-referential type fails because it ends up using _itself_ as a
type bound due to being one of its own active fields.

```rust
#[derive(Reflect)]
struct Foo {
  foo: Vec<Foo>
}

// Foo where Vec<Foo>: Reflect -> Vec<T> where T: Reflect -> Foo where Vec<Foo>: Reflect -> ...
```

## Solution

We can't simply parse all field types for the name of our type. That
would be both complex and prone to errors and false-positives. And even
if it wasn't, what would we replace the bound with?

Instead, I opted to go for a solution that only adds the bounds to what
really needs it: the type parameters. While the bounds on concrete types
make errors a bit cleaner, they aren't strictly necessary. This means we
can change our generated where-clause to only add bounds to generic type
parameters.

Doing this, though, returns us back to the problem of over-bounding
parameters that don't need to be bounded. To solve this, I added a new
container attribute (based on
[this](dtolnay/syn#422 (comment))
comment and @nicopap's
[comment](bevyengine#9046 (comment)))
that allows us to pass in a custom where clause to modify what bounds
are added to these type parameters.

This allows us to do stuff like:

```rust
trait Trait {
  type Assoc;
}

// We don't need `T` to be reflectable since we only care about `T::Assoc`.
#[derive(Reflect)]
#[reflect(where T::Assoc: FromReflect)]
struct Foo<T: Trait>(T::Assoc);

#[derive(TypePath)]
struct Bar;

impl Trait for Bar {
  type Assoc = usize;
}

#[derive(Reflect)]
struct Baz {
  a: Foo<Bar>,
}
```

> **Note**
> I also
[tried](bevyengine@dc139ea)
allowing `#[reflect(ignore)]` to be used on the type parameters
themselves, but that proved problematic since the derive macro does not
consume the attribute. This is why I went with the container attribute
approach.

### Alternatives

One alternative could possibly be to just not add reflection bounds
automatically (i.e. only add required bounds like `Send`, `Sync`, `Any`,
and `TypePath`).

The downside here is we add more friction to using reflection, which
already comes with its own set of considerations. This is a potentially
viable option, but we really need to consider whether or not the
ergonomics hit is worth it.

If we did decide to go the more manual route, we should at least
consider something like bevyengine#5772 to make it easier for users to add the
right bounds (although, this could still become tricky with
`FromReflect` also being automatically derived).

### Open Questions

1. Should we go with this approach or the manual alternative?
2. ~~Should we add a `skip_params` attribute to avoid the `T: 'static`
trick?~~ ~~Decided to go with `custom_where()` as it's the simplest~~
Scratch that, went with a normal where clause
3. ~~`custom_where` bikeshedding?~~ No longer needed since we are using
a normal where clause

### TODO

- [x] Add compile-fail tests

---

## Changelog

- Fixed issue preventing recursive types from deriving `Reflect`
- Changed how where-clause bounds are generated by the `Reflect` derive
macro
- They are now only applied to the type parameters, not to all active
fields
- Added `#[reflect(where T: Trait, U::Assoc: Trait, ...)]` container
attribute

## Migration Guide

When deriving `Reflect`, generic type params that do not need the
automatic reflection bounds (such as `Reflect`) applied to them will
need to opt-out using a custom where clause like: `#[reflect(where T:
Trait, U::Assoc: Trait, ...)]`.

The attribute can define custom bounds only used by the reflection
impls. To simply opt-out all the type params, we can pass in an empty
where clause: `#[reflect(where)]`.

```rust
// BEFORE:
#[derive(Reflect)]
struct Foo<T>(#[reflect(ignore)] T);

// AFTER:
#[derive(Reflect)]
#[reflect(where)]
struct Foo<T>(#[reflect(ignore)] T);
```

---------

Co-authored-by: Nicola Papale <[email protected]>

# Objective

Resolves #4154

Currently, registration must all be done manually:

```rust
#[derive(Reflect)]
struct Foo(Bar);

#[derive(Reflect)]
struct Bar(Baz);

#[derive(Reflect)]
struct Baz(usize);

fn main() {
  // ...
  app
    .register_type::<Foo>()
    .register_type::<Bar>()
    .register_type::<Baz>()
    // .register_type::<usize>() <- This one is handled by Bevy, thankfully
  // ...
}
```

This can grow really quickly and become very annoying to add, remove,
and update as types change. It would be great if we could help reduce
the number of types that a user must manually implement themselves.

## Solution

As suggested in #4154, this PR adds automatic recursive registration.
Essentially, when a type is registered, it may now also choose to
register additional types along with it using the new
`GetTypeRegistration::register_type_dependencies` trait method.

The `Reflect` derive macro now automatically does this for all fields in
structs, tuple structs, struct variants, and tuple variants. This is
also done for tuples, arrays, `Vec<T>`, `HashMap<K, V>`, and
`Option<T>`.

This allows us to simplify the code above like:

```rust
#[derive(Reflect)]
struct Foo(Bar);

#[derive(Reflect)]
struct Bar(Baz);

#[derive(Reflect)]
struct Baz(usize);

fn main() {
  // ...
  app.register_type::<Foo>()
  // ...
}
```

This automatic registration only occurs if the type has not yet been
registered. If it has been registered, we simply skip it and move to the
next one. This reduces the cost of registration and prevents overwriting
customized registrations.

## Considerations

While this does improve ergonomics on one front, it's important to look
at some of the arguments against adopting a PR like this.

#### Generic Bounds

~~Since we need to be able to register the fields individually, we need
those fields to implement `GetTypeRegistration`. This forces users to
then add this trait as a bound on their generic arguments. This
annoyance could be relieved with something like #5772.~~

This is no longer a major issue as the `Reflect` derive now adds the
`GetTypeRegistration` bound by default. This should technically be okay,
since we already add the `Reflect` bound.

However, this can also be considered a breaking change for manual
implementations that left out a `GetTypeRegistration` impl ~~or for
items that contain dynamic types (e.g. `DynamicStruct`) since those also
do not implement `GetTypeRegistration`~~.

#### Registration Assumptions

By automatically registering fields, users might inadvertently be
relying on certain types to be automatically registered. If `Foo`
auto-registers `Bar`, but `Foo` is later removed from the code, then
anywhere that previously used or relied on `Bar`'s registration would
now fail.

---

## Changelog

- Added recursive type registration to structs, tuple structs, struct
variants, tuple variants, tuples, arrays, `Vec<T>`, `HashMap<K, V>`, and
`Option<T>`
- Added a new trait in the hidden `bevy_reflect::__macro_exports` module
called `RegisterForReflection`
- Added `GetTypeRegistration` impl for
`bevy_render::render_asset::RenderAssetUsages`

## Migration Guide

All types that derive `Reflect` will now automatically add
`GetTypeRegistration` as a bound on all (unignored) fields. This means
that all reflected fields will need to also implement
`GetTypeRegistration`.

If all fields **derive** `Reflect` or are implemented in `bevy_reflect`,
this should not cause any issues. However, manual implementations of
`Reflect` that excluded a `GetTypeRegistration` impl for their type will
need to add one.

```rust
#[derive(Reflect)]
struct Foo<T: FromReflect> {
  data: MyCustomType<T>
}

// OLD
impl<T: FromReflect> Reflect for MyCustomType<T> {/* ... */}

// NEW
impl<T: FromReflect + GetTypeRegistration> Reflect for MyCustomType<T> {/* ... */}
impl<T: FromReflect + GetTypeRegistration> GetTypeRegistration for MyCustomType<T> {/* ... */}
```

---------

Co-authored-by: James Liu <[email protected]>
Co-authored-by: radiish <[email protected]>
Co-authored-by: Carter Anderson <[email protected]>

# Objective

Resolves bevyengine#4154

Currently, registration must all be done manually:

```rust
#[derive(Reflect)]
struct Foo(Bar);

#[derive(Reflect)]
struct Bar(Baz);

#[derive(Reflect)]
struct Baz(usize);

fn main() {
  // ...
  app
    .register_type::<Foo>()
    .register_type::<Bar>()
    .register_type::<Baz>()
    // .register_type::<usize>() <- This one is handled by Bevy, thankfully
  // ...
}
```

This can grow really quickly and become very annoying to add, remove,
and update as types change. It would be great if we could help reduce
the number of types that a user must manually implement themselves.

## Solution

As suggested in bevyengine#4154, this PR adds automatic recursive registration.
Essentially, when a type is registered, it may now also choose to
register additional types along with it using the new
`GetTypeRegistration::register_type_dependencies` trait method.

The `Reflect` derive macro now automatically does this for all fields in
structs, tuple structs, struct variants, and tuple variants. This is
also done for tuples, arrays, `Vec<T>`, `HashMap<K, V>`, and
`Option<T>`.

This allows us to simplify the code above like:

```rust
#[derive(Reflect)]
struct Foo(Bar);

#[derive(Reflect)]
struct Bar(Baz);

#[derive(Reflect)]
struct Baz(usize);

fn main() {
  // ...
  app.register_type::<Foo>()
  // ...
}
```

This automatic registration only occurs if the type has not yet been
registered. If it has been registered, we simply skip it and move to the
next one. This reduces the cost of registration and prevents overwriting
customized registrations.

## Considerations

While this does improve ergonomics on one front, it's important to look
at some of the arguments against adopting a PR like this.

#### Generic Bounds

~~Since we need to be able to register the fields individually, we need
those fields to implement `GetTypeRegistration`. This forces users to
then add this trait as a bound on their generic arguments. This
annoyance could be relieved with something like bevyengine#5772.~~

This is no longer a major issue as the `Reflect` derive now adds the
`GetTypeRegistration` bound by default. This should technically be okay,
since we already add the `Reflect` bound.

However, this can also be considered a breaking change for manual
implementations that left out a `GetTypeRegistration` impl ~~or for
items that contain dynamic types (e.g. `DynamicStruct`) since those also
do not implement `GetTypeRegistration`~~.

#### Registration Assumptions

By automatically registering fields, users might inadvertently be
relying on certain types to be automatically registered. If `Foo`
auto-registers `Bar`, but `Foo` is later removed from the code, then
anywhere that previously used or relied on `Bar`'s registration would
now fail.

---

## Changelog

- Added recursive type registration to structs, tuple structs, struct
variants, tuple variants, tuples, arrays, `Vec<T>`, `HashMap<K, V>`, and
`Option<T>`
- Added a new trait in the hidden `bevy_reflect::__macro_exports` module
called `RegisterForReflection`
- Added `GetTypeRegistration` impl for
`bevy_render::render_asset::RenderAssetUsages`

## Migration Guide

All types that derive `Reflect` will now automatically add
`GetTypeRegistration` as a bound on all (unignored) fields. This means
that all reflected fields will need to also implement
`GetTypeRegistration`.

If all fields **derive** `Reflect` or are implemented in `bevy_reflect`,
this should not cause any issues. However, manual implementations of
`Reflect` that excluded a `GetTypeRegistration` impl for their type will
need to add one.

```rust
#[derive(Reflect)]
struct Foo<T: FromReflect> {
  data: MyCustomType<T>
}

// OLD
impl<T: FromReflect> Reflect for MyCustomType<T> {/* ... */}

// NEW
impl<T: FromReflect + GetTypeRegistration> Reflect for MyCustomType<T> {/* ... */}
impl<T: FromReflect + GetTypeRegistration> GetTypeRegistration for MyCustomType<T> {/* ... */}
```

---------

Co-authored-by: James Liu <[email protected]>
Co-authored-by: radiish <[email protected]>
Co-authored-by: Carter Anderson <[email protected]>

# Objective

Resolves bevyengine#4154

Currently, registration must all be done manually:

```rust
#[derive(Reflect)]
struct Foo(Bar);

#[derive(Reflect)]
struct Bar(Baz);

#[derive(Reflect)]
struct Baz(usize);

fn main() {
  // ...
  app
    .register_type::<Foo>()
    .register_type::<Bar>()
    .register_type::<Baz>()
    // .register_type::<usize>() <- This one is handled by Bevy, thankfully
  // ...
}
```

This can grow really quickly and become very annoying to add, remove,
and update as types change. It would be great if we could help reduce
the number of types that a user must manually implement themselves.

## Solution

As suggested in bevyengine#4154, this PR adds automatic recursive registration.
Essentially, when a type is registered, it may now also choose to
register additional types along with it using the new
`GetTypeRegistration::register_type_dependencies` trait method.

The `Reflect` derive macro now automatically does this for all fields in
structs, tuple structs, struct variants, and tuple variants. This is
also done for tuples, arrays, `Vec<T>`, `HashMap<K, V>`, and
`Option<T>`.

This allows us to simplify the code above like:

```rust
#[derive(Reflect)]
struct Foo(Bar);

#[derive(Reflect)]
struct Bar(Baz);

#[derive(Reflect)]
struct Baz(usize);

fn main() {
  // ...
  app.register_type::<Foo>()
  // ...
}
```

This automatic registration only occurs if the type has not yet been
registered. If it has been registered, we simply skip it and move to the
next one. This reduces the cost of registration and prevents overwriting
customized registrations.

## Considerations

While this does improve ergonomics on one front, it's important to look
at some of the arguments against adopting a PR like this.

#### Generic Bounds

~~Since we need to be able to register the fields individually, we need
those fields to implement `GetTypeRegistration`. This forces users to
then add this trait as a bound on their generic arguments. This
annoyance could be relieved with something like bevyengine#5772.~~

This is no longer a major issue as the `Reflect` derive now adds the
`GetTypeRegistration` bound by default. This should technically be okay,
since we already add the `Reflect` bound.

However, this can also be considered a breaking change for manual
implementations that left out a `GetTypeRegistration` impl ~~or for
items that contain dynamic types (e.g. `DynamicStruct`) since those also
do not implement `GetTypeRegistration`~~.

#### Registration Assumptions

By automatically registering fields, users might inadvertently be
relying on certain types to be automatically registered. If `Foo`
auto-registers `Bar`, but `Foo` is later removed from the code, then
anywhere that previously used or relied on `Bar`'s registration would
now fail.

---

## Changelog

- Added recursive type registration to structs, tuple structs, struct
variants, tuple variants, tuples, arrays, `Vec<T>`, `HashMap<K, V>`, and
`Option<T>`
- Added a new trait in the hidden `bevy_reflect::__macro_exports` module
called `RegisterForReflection`
- Added `GetTypeRegistration` impl for
`bevy_render::render_asset::RenderAssetUsages`

## Migration Guide

All types that derive `Reflect` will now automatically add
`GetTypeRegistration` as a bound on all (unignored) fields. This means
that all reflected fields will need to also implement
`GetTypeRegistration`.

If all fields **derive** `Reflect` or are implemented in `bevy_reflect`,
this should not cause any issues. However, manual implementations of
`Reflect` that excluded a `GetTypeRegistration` impl for their type will
need to add one.

```rust
#[derive(Reflect)]
struct Foo<T: FromReflect> {
  data: MyCustomType<T>
}

// OLD
impl<T: FromReflect> Reflect for MyCustomType<T> {/* ... */}

// NEW
impl<T: FromReflect + GetTypeRegistration> Reflect for MyCustomType<T> {/* ... */}
impl<T: FromReflect + GetTypeRegistration> GetTypeRegistration for MyCustomType<T> {/* ... */}
```

---------

Co-authored-by: James Liu <[email protected]>
Co-authored-by: radiish <[email protected]>
Co-authored-by: Carter Anderson <[email protected]>

# Objective

- Implements the [Unique Reflect
RFC](https://github.com/nicopap/rfcs/blob/bevy-reflect-api/rfcs/56-better-reflect.md).

## Solution

- Implements the RFC.
- This implementation differs in some ways from the RFC:
- In the RFC, it was suggested `Reflect: Any` but `PartialReflect:
?Any`. During initial implementation I tried this, but we assume the
`PartialReflect: 'static` in a lot of places and the changes required
crept out of the scope of this PR.
- `PartialReflect::try_into_reflect` originally returned `Option<Box<dyn
Reflect>>` but i changed this to `Result<Box<dyn Reflect>, Box<dyn
PartialReflect>>` since the method takes by value and otherwise there
would be no way to recover the type. `as_full` and `as_full_mut` both
still return `Option<&(mut) dyn Reflect>`.

---

## Changelog

- Added `PartialReflect`.
- `Reflect` is now a subtrait of `PartialReflect`.
- Moved most methods on `Reflect` to the new `PartialReflect`.
- Added `PartialReflect::{as_partial_reflect, as_partial_reflect_mut,
into_partial_reflect}`.
- Added `PartialReflect::{try_as_reflect, try_as_reflect_mut,
try_into_reflect}`.
- Added `<dyn PartialReflect>::{try_downcast_ref, try_downcast_mut,
try_downcast, try_take}` supplementing the methods on `dyn Reflect`.

## Migration Guide

- Most instances of `dyn Reflect` should be changed to `dyn
PartialReflect` which is less restrictive, however trait bounds should
generally stay as `T: Reflect`.
- The new `PartialReflect::{as_partial_reflect, as_partial_reflect_mut,
into_partial_reflect, try_as_reflect, try_as_reflect_mut,
try_into_reflect}` methods as well as `Reflect::{as_reflect,
as_reflect_mut, into_reflect}` will need to be implemented for manual
implementors of `Reflect`.

## Future Work

- This PR is designed to be followed up by another "Unique Reflect Phase
2" that addresses the following points:
- Investigate making serialization revolve around `Reflect` instead of
`PartialReflect`.
- [Remove the `try_*` methods on `dyn PartialReflect` since they are
stop
gaps](#7207 (comment)).
- Investigate usages like `ReflectComponent`. In the places they
currently use `PartialReflect`, should they be changed to use `Reflect`?
- Merging this opens the door to lots of reflection features we haven't
been able to implement.
- We could re-add [the `Reflectable`
trait](https://github.com/bevyengine/bevy/blob/8e3488c88065a94a4f72199587e59341c9b6553d/crates/bevy_reflect/src/reflect.rs#L337-L342)
and make `FromReflect` a requirement to improve [`FromReflect`
ergonomics](bevyengine/rfcs#59). This is
currently not possible because dynamic types cannot sensibly be
`FromReflect`.
  - Since this is an alternative to #5772, #5781 would be made cleaner.

---------

Co-authored-by: Alice Cecile <[email protected]>
Co-authored-by: Gino Valente <[email protected]>