bevy/examples/reflection/function_reflection.rs

221 lines
10 KiB
Rust
Raw Normal View History

bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
//! This example demonstrates how functions can be called dynamically using reflection.
//!
//! Function reflection is useful for calling regular Rust functions in a dynamic context,
//! where the types of arguments, return values, and even the function itself aren't known at compile time.
//!
//! This can be used for things like adding scripting support to your application,
//! processing deserialized reflection data, or even just storing type-erased versions of your functions.
use bevy::reflect::{
func::{
bevy_reflect: Function Overloading (Generic & Variadic Functions) (#15074) # Objective Currently function reflection requires users to manually monomorphize their generic functions. For example: ```rust fn add<T: Add<Output=T>>(a: T, b: T) -> T { a + b } // We have to specify the type of `T`: let reflect_add = add::<i32>.into_function(); ``` This PR doesn't aim to solve that problem—this is just a limitation in Rust. However, it also means that reflected functions can only ever work for a single monomorphization. If we wanted to support other types for `T`, we'd have to create a separate function for each one: ```rust let reflect_add_i32 = add::<i32>.into_function(); let reflect_add_u32 = add::<u32>.into_function(); let reflect_add_f32 = add::<f32>.into_function(); // ... ``` So in addition to requiring manual monomorphization, we also lose the benefit of having a single function handle multiple argument types. If a user wanted to create a small modding script that utilized function reflection, they'd have to either: - Store all sets of supported monomorphizations and require users to call the correct one - Write out some logic to find the correct function based on the given arguments While the first option would work, it wouldn't be very ergonomic. The second option is better, but it adds additional complexity to the user's logic—complexity that `bevy_reflect` could instead take on. ## Solution Introduce [function overloading](https://en.wikipedia.org/wiki/Function_overloading). A `DynamicFunction` can now be overloaded with other `DynamicFunction`s. We can rewrite the above code like so: ```rust let reflect_add = add::<i32> .into_function() .with_overload(add::<u32>) .with_overload(add::<f32>); ``` When invoked, the `DynamicFunction` will attempt to find a matching overload for the given set of arguments. And while I went into this PR only looking to improve generic function reflection, I accidentally added support for variadic functions as well (hence why I use the broader term "overload" over "generic"). ```rust // Supports 1 to 4 arguments let multiply_all = (|a: i32| a) .into_function() .with_overload(|a: i32, b: i32| a * b) .with_overload(|a: i32, b: i32, c: i32| a * b * c) .with_overload(|a: i32, b: i32, c: i32, d: i32| a * b * c * d); ``` This is simply an added bonus to this particular implementation. ~~Full variadic support (i.e. allowing for an indefinite number of arguments) will be added in a later PR.~~ I actually decided to limit the maximum number of arguments to 63 to supplement faster lookups, a reduced memory footprint, and faster cloning. ### Alternatives & Rationale I explored a few options for handling generic functions. This PR is the one I feel the most confident in, but I feel I should mention the others and why I ultimately didn't move forward with them. #### Adding `GenericDynamicFunction` **TL;DR:** Adding a distinct `GenericDynamicFunction` type unnecessarily splits and complicates the API. <details> <summary>Details</summary> My initial explorations involved a dedicated `GenericDynamicFunction` to contain and handle the mappings. This was initially started back when `DynamicFunction` was distinct from `DynamicClosure`. My goal was to not prevent us from being able to somehow make `DynamicFunction` implement `Copy`. But once we reverted back to a single `DynamicFunction`, that became a non-issue. But that aside, the real problem was that it created a split in the API. If I'm using a third-party library that uses function reflection, I have to know whether to request a `DynamicFunction` or a `GenericDynamicFunction`. I might not even know ahead of time which one I want. It might need to be determined at runtime. And if I'm creating a library, I might want a type to contain both `DynamicFunction` and `GenericDynamicFunction`. This might not be possible if, for example, I need to store the function in a `HashMap`. The other concern is with `IntoFunction`. Right now `DynamicFunction` trivially implements `IntoFunction` since it can just return itself. But what should `GenericDynamicFunction` do? It could return itself wrapped into a `DynamicFunction`, but then the API for `DynamicFunction` would have to account for this. So then what was the point of having a separate `GenericDynamicFunction` anyways? And even apart from `IntoFunction`, there's nothing stopping someone from manually creating a generic `DynamicFunction` through lying about its `FunctionInfo` and wrapping a `GenericDynamicFunction`. That being said, this is probably the "best" alternative if we added a `Function` trait and stored functions as `Box<dyn Function>`. However, I'm not convinced we gain much from this. Sure, we could keep the API for `DynamicFunction` the same, but consumers of `Function` will need to account for `GenericDynamicFunction` regardless (e.g. handling multiple `FunctionInfo`, a ranged argument count, etc.). And for all cases, except where using `DynamicFunction` directly, you end up treating them all like `GenericDynamicFunction`. Right now, if we did go with `GenericDynamicFunction`, the only major benefit we'd gain would be saving 24 bytes. If memory ever does become an issue here, we could swap over. But I think for the time being it's better for us to pursue a clearer mental model and end-user ergonomics through unification. </details> ##### Using the `FunctionRegistry` **TL;DR:** Having overloads only exist in the `FunctionRegistry` unnecessarily splits and complicates the API. <details> <summary>Details</summary> Another idea was to store the overloads in the `FunctionRegistry`. Users would then just call functions directly through the registry (i.e. `registry.call("my_func", my_args)`). I didn't go with this option because of how it specifically relies on the functions being registered. You'd not only always need access to the registry, but you'd need to ensure that the functions you want to call are even registered. It also means you can't just store a generic `DynamicFunction` on a type. Instead, you'll need to store the function's name and use that to look up the function in the registry—even if it's only ever used by that type. Doing so also removes all the benefits of `DynamicFunction`, such as the ability to pass it to functions accepting `IntoFunction`, modify it if needed, and so on. Like `GenericDynamicFunction` this introduces a split in the ecosystem: you either store `DynamicFunction`, store a string to look up the function, or force `DynamicFunction` to wrap your generic function anyways. Or worse yet: have `DynamicFunction` wrap the lookup function using `FunctionRegistryArc`. </details> #### Generic `ArgInfo` **TL;DR:** Allowing `ArgInfo` and `ReturnInfo` to store the generic information introduces a footgun when interpreting `FunctionInfo`. <details> <summary>Details</summary> Regardless of how we represent a generic function, one thing is clear: we need to be able to represent the information for such a function. This PR does so by introducing a `FunctionInfoType` enum to wrap one or more `FunctionInfo` values. Originally, I didn't do this. I had `ArgInfo` and `ReturnInfo` allow for generic types. This allowed us to have a single `FunctionInfo` to represent our function, but then I realized that it actually lies about our function. If we have two `ArgInfo` that both allow for either `i32` or `u32`, what does this tell us about our function? It turns out: nothing! We can't know whether our function takes `(i32, i32)`, `(u32, u32)`, `(i32, u32)`, or `(u32, i32)`. It therefore makes more sense to just represent a function with multiple `FunctionInfo` since that's really what it's made up of. </details> #### Flatten `FunctionInfo` **TL;DR:** Flattening removes additional per-overload information some users may desire and prevents us from adding more information in the future. <details> <summary>Details</summary> Why don't we just flatten multiple `FunctionInfo` into just one that can contain multiple signatures? This is something we could do, but I decided against it for a few reasons: - The only thing we'd be able to get rid of for each signature would be the `name`. While not enough to not do it, it doesn't really suggest we *have* to either. - Some consumers may want access to the names of the functions that make up the overloaded function. For example, to track a bug where an undesirable function is being added as an overload. Or to more easily locate the original function of an overload. - We may eventually allow for more information to be stored on `FunctionInfo`. For example, we may allow for documentation to be stored like we do for `TypeInfo`. Consumers of this documentation may want access to the documentation of each overload as they may provide documentation specific to that overload. </details> ## Testing This PR adds lots of tests and benchmarks, and also adds to the example. To run the tests: ``` cargo test --package bevy_reflect --all-features ``` To run the benchmarks: ``` cargo bench --bench reflect_function --all-features ``` To run the example: ``` cargo run --package bevy --example function_reflection --all-features ``` ### Benchmarks One of my goals with this PR was to leave the typical case of non-overloaded functions largely unaffected by the changes introduced in this PR. ~~And while the static size of `DynamicFunction` has increased by 17% (from 136 to 160 bytes), the performance has generally stayed the same~~ The static size of `DynamicFunction` has decreased from 136 to 112 bytes, while calling performance has generally stayed the same: | | `main` | 7d293ab | 252f3897d | |-------------------------------------|--------|---------|-----------| | `into/function` | 37 ns | 46 ns | 142 ns | | `with_overload/01_simple_overload` | - | 149 ns | 268 ns | | `with_overload/01_complex_overload` | - | 332 ns | 431 ns | | `with_overload/10_simple_overload` | - | 1266 ns | 2618 ns | | `with_overload/10_complex_overload` | - | 2544 ns | 4170 ns | | `call/function` | 57 ns | 58 ns | 61 ns | | `call/01_simple_overload` | - | 255 ns | 242 ns | | `call/01_complex_overload` | - | 595 ns | 431 ns | | `call/10_simple_overload` | - | 740 ns | 699 ns | | `call/10_complex_overload` | - | 1824 ns | 1618 ns | For the overloaded function tests, the leading number indicates how many overloads there are: `01` indicates 1 overload, `10` indicates 10 overloads. The `complex` cases have 10 unique generic types and 10 arguments, compared to the `simple` 1 generic type and 2 arguments. I aimed to prioritize the performance of calling the functions over creating them, hence creation speed tends to be a bit slower. There may be other optimizations we can look into but that's probably best saved for a future PR. The important bit is that the standard ~~`into/function`~~ and `call/function` benchmarks show minimal regressions. Since the latest changes, `into/function` does have some regressions, but again the priority was `call/function`. We can probably optimize `into/function` if needed in the future. --- ## Showcase Function reflection now supports [function overloading](https://en.wikipedia.org/wiki/Function_overloading)! This can be used to simulate generic functions: ```rust fn add<T: Add<Output=T>>(a: T, b: T) -> T { a + b } let reflect_add = add::<i32> .into_function() .with_overload(add::<u32>) .with_overload(add::<f32>); let args = ArgList::default().push_owned(25_i32).push_owned(75_i32); let result = func.call(args).unwrap().unwrap_owned(); assert_eq!(result.try_take::<i32>().unwrap(), 100); let args = ArgList::default().push_owned(25.0_f32).push_owned(75.0_f32); let result = func.call(args).unwrap().unwrap_owned(); assert_eq!(result.try_take::<f32>().unwrap(), 100.0); ``` You can also simulate variadic functions: ```rust #[derive(Reflect, PartialEq, Debug)] struct Player { name: Option<String>, health: u32, } // Creates a `Player` with one of the following: // - No name and 100 health // - A name and 100 health // - No name and custom health // - A name and custom health let create_player = (|| Player { name: None, health: 100, }) .into_function() .with_overload(|name: String| Player { name: Some(name), health: 100, }) .with_overload(|health: u32| Player { name: None, health }) .with_overload(|name: String, health: u32| Player { name: Some(name), health, }); let args = ArgList::default() .push_owned(String::from("Urist")) .push_owned(55_u32); let player = create_player .call(args) .unwrap() .unwrap_owned() .try_take::<Player>() .unwrap(); assert_eq!( player, Player { name: Some(String::from("Urist")), health: 55 } ); ```
2024-12-10 01:51:47 +00:00
ArgList, DynamicFunction, DynamicFunctionMut, FunctionResult, IntoFunction,
IntoFunctionMut, Return, SignatureInfo,
},
PartialReflect, Reflect,
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
};
// Note that the `dbg!` invocations are used purely for demonstration purposes
// and are not strictly necessary for the example to work.
fn main() {
// There are times when it may be helpful to store a function away for later.
// In Rust, we can do this by storing either a function pointer or a function trait object.
// For example, say we wanted to store the following function:
fn add(left: i32, right: i32) -> i32 {
left + right
}
// We could store it as either of the following:
let fn_pointer: fn(i32, i32) -> i32 = add;
let fn_trait_object: Box<dyn Fn(i32, i32) -> i32> = Box::new(add);
// And we can call them like so:
let result = fn_pointer(2, 2);
assert_eq!(result, 4);
let result = fn_trait_object(2, 2);
assert_eq!(result, 4);
// However, you'll notice that we have to know the types of the arguments and return value at compile time.
// This means there's not really a way to store or call these functions dynamically at runtime.
// Luckily, Bevy's reflection crate comes with a set of tools for doing just that!
// We do this by first converting our function into the reflection-based `DynamicFunction` type
// using the `IntoFunction` trait.
bevy_reflect: Function reflection terminology refactor (#14813) # Objective One of the changes in #14704 made `DynamicFunction` effectively the same as `DynamicClosure<'static>`. This change meant that the de facto function type would likely be `DynamicClosure<'static>` instead of the intended `DynamicFunction`, since the former is much more flexible. We _could_ explore ways of making `DynamicFunction` implement `Copy` using some unsafe code, but it likely wouldn't be worth it. And users would likely still reach for the convenience of `DynamicClosure<'static>` over the copy-ability of `DynamicFunction`. The goal of this PR is to fix this confusion between the two types. ## Solution Firstly, the `DynamicFunction` type was removed. Again, it was no different than `DynamicClosure<'static>` so it wasn't a huge deal to remove. Secondly, `DynamicClosure<'env>` and `DynamicClosureMut<'env>` were renamed to `DynamicFunction<'env>` and `DynamicFunctionMut<'env>`, respectively. Yes, we still ultimately kept the naming of `DynamicFunction`, but changed its behavior to that of `DynamicClosure<'env>`. We need a term to refer to both functions and closures, and "function" was the best option. [Originally](https://discord.com/channels/691052431525675048/1002362493634629796/1274091992162242710), I was going to go with "callable" as the replacement term to encompass both functions and closures (e.g. `DynamciCallable<'env>`). However, it was [suggested](https://discord.com/channels/691052431525675048/1002362493634629796/1274653581777047625) by @SkiFire13 that the simpler "function" term could be used instead. While "callable" is perhaps the better umbrella term—being truly ambiguous over functions and closures— "function" is more familiar, used more often, easier to discover, and is subjectively just "better-sounding". ## Testing Most changes are purely swapping type names or updating documentation, but you can verify everything still works by running the following command: ``` cargo test --package bevy_reflect ```
2024-08-19 21:52:36 +00:00
let function: DynamicFunction<'static> = dbg!(add.into_function());
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
// This time, you'll notice that `DynamicFunction` doesn't take any information about the function's arguments or return value.
// This is because `DynamicFunction` checks the types of the arguments and return value at runtime.
// Now we can generate a list of arguments:
let args: ArgList = dbg!(ArgList::new().push_owned(2_i32).push_owned(2_i32));
// And finally, we can call the function.
// This returns a `Result` indicating whether the function was called successfully.
// For now, we'll just unwrap it to get our `Return` value,
// which is an enum containing the function's return value.
let return_value: Return = dbg!(function.call(args).unwrap());
// The `Return` value can be pattern matched or unwrapped to get the underlying reflection data.
// For the sake of brevity, we'll just unwrap it here and downcast it to the expected type of `i32`.
reflect: implement the unique reflect rfc (#7207) # 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](https://github.com/bevyengine/bevy/pull/7207#discussion_r1083476050). - 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](https://github.com/bevyengine/rfcs/pull/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 <alice.i.cecile@gmail.com> Co-authored-by: Gino Valente <49806985+MrGVSV@users.noreply.github.com>
2024-08-12 17:01:41 +00:00
let value: Box<dyn PartialReflect> = return_value.unwrap_owned();
assert_eq!(value.try_take::<i32>().unwrap(), 4);
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
bevy_reflect: Add `DynamicClosure` and `DynamicClosureMut` (#14141) # Objective As mentioned in [this](https://github.com/bevyengine/bevy/pull/13152#issuecomment-2198387297) comment, creating a function registry (see #14098) is a bit difficult due to the requirements of `DynamicFunction`. Internally, a `DynamicFunction` contains a `Box<dyn FnMut>` (the function that reifies reflected arguments and calls the actual function), which requires `&mut self` in order to be called. This means that users would require a mutable reference to the function registry for it to be useful— which isn't great. And they can't clone the `DynamicFunction` either because cloning an `FnMut` isn't really feasible (wrapping it in an `Arc` would allow it to be cloned but we wouldn't be able to call the clone since we need a mutable reference to the `FnMut`, which we can't get with multiple `Arc`s still alive, requiring us to also slap in a `Mutex`, which adds additional overhead). And we don't want to just replace the `dyn FnMut` with `dyn Fn` as that would prevent reflecting closures that mutate their environment. Instead, we need to introduce a new type to split the requirements of `DynamicFunction`. ## Solution Introduce new types for representing closures. Specifically, this PR introduces `DynamicClosure` and `DynamicClosureMut`. Similar to how `IntoFunction` exists for `DynamicFunction`, two new traits were introduced: `IntoClosure` and `IntoClosureMut`. Now `DynamicFunction` stores a `dyn Fn` with a `'static` lifetime. `DynamicClosure` also uses a `dyn Fn` but has a lifetime, `'env`, tied to its environment. `DynamicClosureMut` is most like the old `DynamicFunction`, keeping the `dyn FnMut` and also typing its lifetime, `'env`, to the environment Here are some comparison tables: | | `DynamicFunction` | `DynamicClosure` | `DynamicClosureMut` | | - | ----------------- | ---------------- | ------------------- | | Callable with `&self` | ✅ | ✅ | ❌ | | Callable with `&mut self` | ✅ | ✅ | ✅ | | Allows for non-`'static` lifetimes | ❌ | ✅ | ✅ | | | `IntoFunction` | `IntoClosure` | `IntoClosureMut` | | - | -------------- | ------------- | ---------------- | | Convert `fn` functions | ✅ | ✅ | ✅ | | Convert `fn` methods | ✅ | ✅ | ✅ | | Convert anonymous functions | ✅ | ✅ | ✅ | | Convert closures that capture immutable references | ❌ | ✅ | ✅ | | Convert closures that capture mutable references | ❌ | ❌ | ✅ | | Convert closures that capture owned values | ❌[^1] | ✅ | ✅ | [^1]: Due to limitations in Rust, `IntoFunction` can't be implemented for just functions (unless we forced users to manually coerce them to function pointers first). So closures that meet the trait requirements _can technically_ be converted into a `DynamicFunction` as well. To both future-proof and reduce confusion, though, we'll just pretend like this isn't a thing. ```rust let mut list: Vec<i32> = vec![1, 2, 3]; // `replace` is a closure that captures a mutable reference to `list` let mut replace = |index: usize, value: i32| -> i32 { let old_value = list[index]; list[index] = value; old_value }; // Convert the closure into a dynamic closure using `IntoClosureMut::into_closure_mut` let mut func: DynamicClosureMut = replace.into_closure_mut(); // Dynamically call the closure: let args = ArgList::default().push_owned(1_usize).push_owned(-2_i32); let value = func.call_once(args).unwrap().unwrap_owned(); // Check the result: assert_eq!(value.take::<i32>().unwrap(), 2); assert_eq!(list, vec![1, -2, 3]); ``` ### `ReflectFn`/`ReflectFnMut` To make extending the function reflection system easier (the blanket impls for `IntoFunction`, `IntoClosure`, and `IntoClosureMut` are all incredibly short), this PR generalizes callables with two new traits: `ReflectFn` and `ReflectFnMut`. These traits mimic `Fn` and `FnMut` but allow for being called via reflection. In fact, their blanket implementations are identical save for `ReflectFn` being implemented over `Fn` types and `ReflectFnMut` being implemented over `FnMut` types. And just as `Fn` is a subtrait of `FnMut`, `ReflectFn` is a subtrait of `ReflectFnMut`. So anywhere that expects a `ReflectFnMut` can also be given a `ReflectFn`. To reiterate, these traits aren't 100% necessary. They were added in purely for extensibility. If we decide to split things up differently or add new traits/types in the future, then those changes should be much simpler to implement. ### `TypedFunction` Because of the split into `ReflectFn` and `ReflectFnMut`, we needed a new way to access the function type information. This PR moves that concept over into `TypedFunction`. Much like `Typed`, this provides a way to access a function's `FunctionInfo`. By splitting this trait out, it helps to ensure the other traits are focused on a single responsibility. ### Internal Macros The original function PR (#13152) implemented `IntoFunction` using a macro which was passed into an `all_tuples!` macro invocation. Because we needed the same functionality for these new traits, this PR has copy+pasted that code for `ReflectFn`, `ReflectFnMut`, and `TypedFunction`— albeit with some differences between them. Originally, I was going to try and macro-ify the impls and where clauses such that we wouldn't have to straight up duplicate a lot of this logic. However, aside from being more complex in general, autocomplete just does not play nice with such heavily nested macros (tried in both RustRover and VSCode). And both of those problems told me that it just wasn't worth it: we need to ensure the crate is easily maintainable, even at the cost of duplicating code. So instead, I made sure to simplify the macro code by removing all fully-qualified syntax and cutting the where clauses down to the bare essentials, which helps to clean up a lot of the visual noise. I also tried my best to document the macro logic in certain areas (I may even add a bit more) to help with maintainability for future devs. ### Documentation Documentation for this module was a bit difficult for me. So many of these traits and types are very interconnected. And each trait/type has subtle differences that make documenting it in a single place, like at the module level, difficult to do cleanly. Describing the valid signatures is also challenging to do well. Hopefully what I have here is okay. I think I did an okay job, but let me know if there any thoughts on ways to improve it. We can also move such a task to a followup PR for more focused discussion. ## Testing You can test locally by running: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Added `DynamicClosure` struct - Added `DynamicClosureMut` struct - Added `IntoClosure` trait - Added `IntoClosureMut` trait - Added `ReflectFn` trait - Added `ReflectFnMut` trait - Added `TypedFunction` trait - `IntoFunction` now only works for standard Rust functions - `IntoFunction` no longer takes a lifetime parameter - `DynamicFunction::call` now only requires `&self` - Removed `DynamicFunction::call_once` - Changed the `IntoReturn::into_return` signature to include a where clause ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. ### `IntoClosure` `IntoFunction` now only works for standard Rust functions. Calling `IntoFunction::into_function` on a closure that captures references to its environment (either mutable or immutable), will no longer compile. Instead, you will need to use either `IntoClosure::into_closure` to create a `DynamicClosure` or `IntoClosureMut::into_closure_mut` to create a `DynamicClosureMut`, depending on your needs: ```rust let punct = String::from("!"); let print = |value: String| { println!("{value}{punct}"); }; // BEFORE let func: DynamicFunction = print.into_function(); // AFTER let func: DynamicClosure = print.into_closure(); ``` ### `IntoFunction` lifetime Additionally, `IntoFunction` no longer takes a lifetime parameter as it always expects a `'static` lifetime. Usages will need to remove any lifetime parameters: ```rust // BEFORE fn execute<'env, F: IntoFunction<'env, Marker>, Marker>(f: F) {/* ... */} // AFTER fn execute<F: IntoFunction<Marker>, Marker>(f: F) {/* ... */} ``` ### `IntoReturn` `IntoReturn::into_return` now has a where clause. Any manual implementors will need to add this where clause to their implementation.
2024-07-16 03:22:43 +00:00
// The same can also be done for closures that capture references to their environment.
bevy_reflect: Function reflection terminology refactor (#14813) # Objective One of the changes in #14704 made `DynamicFunction` effectively the same as `DynamicClosure<'static>`. This change meant that the de facto function type would likely be `DynamicClosure<'static>` instead of the intended `DynamicFunction`, since the former is much more flexible. We _could_ explore ways of making `DynamicFunction` implement `Copy` using some unsafe code, but it likely wouldn't be worth it. And users would likely still reach for the convenience of `DynamicClosure<'static>` over the copy-ability of `DynamicFunction`. The goal of this PR is to fix this confusion between the two types. ## Solution Firstly, the `DynamicFunction` type was removed. Again, it was no different than `DynamicClosure<'static>` so it wasn't a huge deal to remove. Secondly, `DynamicClosure<'env>` and `DynamicClosureMut<'env>` were renamed to `DynamicFunction<'env>` and `DynamicFunctionMut<'env>`, respectively. Yes, we still ultimately kept the naming of `DynamicFunction`, but changed its behavior to that of `DynamicClosure<'env>`. We need a term to refer to both functions and closures, and "function" was the best option. [Originally](https://discord.com/channels/691052431525675048/1002362493634629796/1274091992162242710), I was going to go with "callable" as the replacement term to encompass both functions and closures (e.g. `DynamciCallable<'env>`). However, it was [suggested](https://discord.com/channels/691052431525675048/1002362493634629796/1274653581777047625) by @SkiFire13 that the simpler "function" term could be used instead. While "callable" is perhaps the better umbrella term—being truly ambiguous over functions and closures— "function" is more familiar, used more often, easier to discover, and is subjectively just "better-sounding". ## Testing Most changes are purely swapping type names or updating documentation, but you can verify everything still works by running the following command: ``` cargo test --package bevy_reflect ```
2024-08-19 21:52:36 +00:00
// Closures that capture their environment immutably can be converted into a `DynamicFunction`
// using the `IntoFunction` trait.
bevy_reflect: Add `DynamicClosure` and `DynamicClosureMut` (#14141) # Objective As mentioned in [this](https://github.com/bevyengine/bevy/pull/13152#issuecomment-2198387297) comment, creating a function registry (see #14098) is a bit difficult due to the requirements of `DynamicFunction`. Internally, a `DynamicFunction` contains a `Box<dyn FnMut>` (the function that reifies reflected arguments and calls the actual function), which requires `&mut self` in order to be called. This means that users would require a mutable reference to the function registry for it to be useful— which isn't great. And they can't clone the `DynamicFunction` either because cloning an `FnMut` isn't really feasible (wrapping it in an `Arc` would allow it to be cloned but we wouldn't be able to call the clone since we need a mutable reference to the `FnMut`, which we can't get with multiple `Arc`s still alive, requiring us to also slap in a `Mutex`, which adds additional overhead). And we don't want to just replace the `dyn FnMut` with `dyn Fn` as that would prevent reflecting closures that mutate their environment. Instead, we need to introduce a new type to split the requirements of `DynamicFunction`. ## Solution Introduce new types for representing closures. Specifically, this PR introduces `DynamicClosure` and `DynamicClosureMut`. Similar to how `IntoFunction` exists for `DynamicFunction`, two new traits were introduced: `IntoClosure` and `IntoClosureMut`. Now `DynamicFunction` stores a `dyn Fn` with a `'static` lifetime. `DynamicClosure` also uses a `dyn Fn` but has a lifetime, `'env`, tied to its environment. `DynamicClosureMut` is most like the old `DynamicFunction`, keeping the `dyn FnMut` and also typing its lifetime, `'env`, to the environment Here are some comparison tables: | | `DynamicFunction` | `DynamicClosure` | `DynamicClosureMut` | | - | ----------------- | ---------------- | ------------------- | | Callable with `&self` | ✅ | ✅ | ❌ | | Callable with `&mut self` | ✅ | ✅ | ✅ | | Allows for non-`'static` lifetimes | ❌ | ✅ | ✅ | | | `IntoFunction` | `IntoClosure` | `IntoClosureMut` | | - | -------------- | ------------- | ---------------- | | Convert `fn` functions | ✅ | ✅ | ✅ | | Convert `fn` methods | ✅ | ✅ | ✅ | | Convert anonymous functions | ✅ | ✅ | ✅ | | Convert closures that capture immutable references | ❌ | ✅ | ✅ | | Convert closures that capture mutable references | ❌ | ❌ | ✅ | | Convert closures that capture owned values | ❌[^1] | ✅ | ✅ | [^1]: Due to limitations in Rust, `IntoFunction` can't be implemented for just functions (unless we forced users to manually coerce them to function pointers first). So closures that meet the trait requirements _can technically_ be converted into a `DynamicFunction` as well. To both future-proof and reduce confusion, though, we'll just pretend like this isn't a thing. ```rust let mut list: Vec<i32> = vec![1, 2, 3]; // `replace` is a closure that captures a mutable reference to `list` let mut replace = |index: usize, value: i32| -> i32 { let old_value = list[index]; list[index] = value; old_value }; // Convert the closure into a dynamic closure using `IntoClosureMut::into_closure_mut` let mut func: DynamicClosureMut = replace.into_closure_mut(); // Dynamically call the closure: let args = ArgList::default().push_owned(1_usize).push_owned(-2_i32); let value = func.call_once(args).unwrap().unwrap_owned(); // Check the result: assert_eq!(value.take::<i32>().unwrap(), 2); assert_eq!(list, vec![1, -2, 3]); ``` ### `ReflectFn`/`ReflectFnMut` To make extending the function reflection system easier (the blanket impls for `IntoFunction`, `IntoClosure`, and `IntoClosureMut` are all incredibly short), this PR generalizes callables with two new traits: `ReflectFn` and `ReflectFnMut`. These traits mimic `Fn` and `FnMut` but allow for being called via reflection. In fact, their blanket implementations are identical save for `ReflectFn` being implemented over `Fn` types and `ReflectFnMut` being implemented over `FnMut` types. And just as `Fn` is a subtrait of `FnMut`, `ReflectFn` is a subtrait of `ReflectFnMut`. So anywhere that expects a `ReflectFnMut` can also be given a `ReflectFn`. To reiterate, these traits aren't 100% necessary. They were added in purely for extensibility. If we decide to split things up differently or add new traits/types in the future, then those changes should be much simpler to implement. ### `TypedFunction` Because of the split into `ReflectFn` and `ReflectFnMut`, we needed a new way to access the function type information. This PR moves that concept over into `TypedFunction`. Much like `Typed`, this provides a way to access a function's `FunctionInfo`. By splitting this trait out, it helps to ensure the other traits are focused on a single responsibility. ### Internal Macros The original function PR (#13152) implemented `IntoFunction` using a macro which was passed into an `all_tuples!` macro invocation. Because we needed the same functionality for these new traits, this PR has copy+pasted that code for `ReflectFn`, `ReflectFnMut`, and `TypedFunction`— albeit with some differences between them. Originally, I was going to try and macro-ify the impls and where clauses such that we wouldn't have to straight up duplicate a lot of this logic. However, aside from being more complex in general, autocomplete just does not play nice with such heavily nested macros (tried in both RustRover and VSCode). And both of those problems told me that it just wasn't worth it: we need to ensure the crate is easily maintainable, even at the cost of duplicating code. So instead, I made sure to simplify the macro code by removing all fully-qualified syntax and cutting the where clauses down to the bare essentials, which helps to clean up a lot of the visual noise. I also tried my best to document the macro logic in certain areas (I may even add a bit more) to help with maintainability for future devs. ### Documentation Documentation for this module was a bit difficult for me. So many of these traits and types are very interconnected. And each trait/type has subtle differences that make documenting it in a single place, like at the module level, difficult to do cleanly. Describing the valid signatures is also challenging to do well. Hopefully what I have here is okay. I think I did an okay job, but let me know if there any thoughts on ways to improve it. We can also move such a task to a followup PR for more focused discussion. ## Testing You can test locally by running: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Added `DynamicClosure` struct - Added `DynamicClosureMut` struct - Added `IntoClosure` trait - Added `IntoClosureMut` trait - Added `ReflectFn` trait - Added `ReflectFnMut` trait - Added `TypedFunction` trait - `IntoFunction` now only works for standard Rust functions - `IntoFunction` no longer takes a lifetime parameter - `DynamicFunction::call` now only requires `&self` - Removed `DynamicFunction::call_once` - Changed the `IntoReturn::into_return` signature to include a where clause ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. ### `IntoClosure` `IntoFunction` now only works for standard Rust functions. Calling `IntoFunction::into_function` on a closure that captures references to its environment (either mutable or immutable), will no longer compile. Instead, you will need to use either `IntoClosure::into_closure` to create a `DynamicClosure` or `IntoClosureMut::into_closure_mut` to create a `DynamicClosureMut`, depending on your needs: ```rust let punct = String::from("!"); let print = |value: String| { println!("{value}{punct}"); }; // BEFORE let func: DynamicFunction = print.into_function(); // AFTER let func: DynamicClosure = print.into_closure(); ``` ### `IntoFunction` lifetime Additionally, `IntoFunction` no longer takes a lifetime parameter as it always expects a `'static` lifetime. Usages will need to remove any lifetime parameters: ```rust // BEFORE fn execute<'env, F: IntoFunction<'env, Marker>, Marker>(f: F) {/* ... */} // AFTER fn execute<F: IntoFunction<Marker>, Marker>(f: F) {/* ... */} ``` ### `IntoReturn` `IntoReturn::into_return` now has a where clause. Any manual implementors will need to add this where clause to their implementation.
2024-07-16 03:22:43 +00:00
let minimum = 5;
let clamp = |value: i32| value.max(minimum);
bevy_reflect: Function reflection terminology refactor (#14813) # Objective One of the changes in #14704 made `DynamicFunction` effectively the same as `DynamicClosure<'static>`. This change meant that the de facto function type would likely be `DynamicClosure<'static>` instead of the intended `DynamicFunction`, since the former is much more flexible. We _could_ explore ways of making `DynamicFunction` implement `Copy` using some unsafe code, but it likely wouldn't be worth it. And users would likely still reach for the convenience of `DynamicClosure<'static>` over the copy-ability of `DynamicFunction`. The goal of this PR is to fix this confusion between the two types. ## Solution Firstly, the `DynamicFunction` type was removed. Again, it was no different than `DynamicClosure<'static>` so it wasn't a huge deal to remove. Secondly, `DynamicClosure<'env>` and `DynamicClosureMut<'env>` were renamed to `DynamicFunction<'env>` and `DynamicFunctionMut<'env>`, respectively. Yes, we still ultimately kept the naming of `DynamicFunction`, but changed its behavior to that of `DynamicClosure<'env>`. We need a term to refer to both functions and closures, and "function" was the best option. [Originally](https://discord.com/channels/691052431525675048/1002362493634629796/1274091992162242710), I was going to go with "callable" as the replacement term to encompass both functions and closures (e.g. `DynamciCallable<'env>`). However, it was [suggested](https://discord.com/channels/691052431525675048/1002362493634629796/1274653581777047625) by @SkiFire13 that the simpler "function" term could be used instead. While "callable" is perhaps the better umbrella term—being truly ambiguous over functions and closures— "function" is more familiar, used more often, easier to discover, and is subjectively just "better-sounding". ## Testing Most changes are purely swapping type names or updating documentation, but you can verify everything still works by running the following command: ``` cargo test --package bevy_reflect ```
2024-08-19 21:52:36 +00:00
let function: DynamicFunction = dbg!(clamp.into_function());
bevy_reflect: Add `DynamicClosure` and `DynamicClosureMut` (#14141) # Objective As mentioned in [this](https://github.com/bevyengine/bevy/pull/13152#issuecomment-2198387297) comment, creating a function registry (see #14098) is a bit difficult due to the requirements of `DynamicFunction`. Internally, a `DynamicFunction` contains a `Box<dyn FnMut>` (the function that reifies reflected arguments and calls the actual function), which requires `&mut self` in order to be called. This means that users would require a mutable reference to the function registry for it to be useful— which isn't great. And they can't clone the `DynamicFunction` either because cloning an `FnMut` isn't really feasible (wrapping it in an `Arc` would allow it to be cloned but we wouldn't be able to call the clone since we need a mutable reference to the `FnMut`, which we can't get with multiple `Arc`s still alive, requiring us to also slap in a `Mutex`, which adds additional overhead). And we don't want to just replace the `dyn FnMut` with `dyn Fn` as that would prevent reflecting closures that mutate their environment. Instead, we need to introduce a new type to split the requirements of `DynamicFunction`. ## Solution Introduce new types for representing closures. Specifically, this PR introduces `DynamicClosure` and `DynamicClosureMut`. Similar to how `IntoFunction` exists for `DynamicFunction`, two new traits were introduced: `IntoClosure` and `IntoClosureMut`. Now `DynamicFunction` stores a `dyn Fn` with a `'static` lifetime. `DynamicClosure` also uses a `dyn Fn` but has a lifetime, `'env`, tied to its environment. `DynamicClosureMut` is most like the old `DynamicFunction`, keeping the `dyn FnMut` and also typing its lifetime, `'env`, to the environment Here are some comparison tables: | | `DynamicFunction` | `DynamicClosure` | `DynamicClosureMut` | | - | ----------------- | ---------------- | ------------------- | | Callable with `&self` | ✅ | ✅ | ❌ | | Callable with `&mut self` | ✅ | ✅ | ✅ | | Allows for non-`'static` lifetimes | ❌ | ✅ | ✅ | | | `IntoFunction` | `IntoClosure` | `IntoClosureMut` | | - | -------------- | ------------- | ---------------- | | Convert `fn` functions | ✅ | ✅ | ✅ | | Convert `fn` methods | ✅ | ✅ | ✅ | | Convert anonymous functions | ✅ | ✅ | ✅ | | Convert closures that capture immutable references | ❌ | ✅ | ✅ | | Convert closures that capture mutable references | ❌ | ❌ | ✅ | | Convert closures that capture owned values | ❌[^1] | ✅ | ✅ | [^1]: Due to limitations in Rust, `IntoFunction` can't be implemented for just functions (unless we forced users to manually coerce them to function pointers first). So closures that meet the trait requirements _can technically_ be converted into a `DynamicFunction` as well. To both future-proof and reduce confusion, though, we'll just pretend like this isn't a thing. ```rust let mut list: Vec<i32> = vec![1, 2, 3]; // `replace` is a closure that captures a mutable reference to `list` let mut replace = |index: usize, value: i32| -> i32 { let old_value = list[index]; list[index] = value; old_value }; // Convert the closure into a dynamic closure using `IntoClosureMut::into_closure_mut` let mut func: DynamicClosureMut = replace.into_closure_mut(); // Dynamically call the closure: let args = ArgList::default().push_owned(1_usize).push_owned(-2_i32); let value = func.call_once(args).unwrap().unwrap_owned(); // Check the result: assert_eq!(value.take::<i32>().unwrap(), 2); assert_eq!(list, vec![1, -2, 3]); ``` ### `ReflectFn`/`ReflectFnMut` To make extending the function reflection system easier (the blanket impls for `IntoFunction`, `IntoClosure`, and `IntoClosureMut` are all incredibly short), this PR generalizes callables with two new traits: `ReflectFn` and `ReflectFnMut`. These traits mimic `Fn` and `FnMut` but allow for being called via reflection. In fact, their blanket implementations are identical save for `ReflectFn` being implemented over `Fn` types and `ReflectFnMut` being implemented over `FnMut` types. And just as `Fn` is a subtrait of `FnMut`, `ReflectFn` is a subtrait of `ReflectFnMut`. So anywhere that expects a `ReflectFnMut` can also be given a `ReflectFn`. To reiterate, these traits aren't 100% necessary. They were added in purely for extensibility. If we decide to split things up differently or add new traits/types in the future, then those changes should be much simpler to implement. ### `TypedFunction` Because of the split into `ReflectFn` and `ReflectFnMut`, we needed a new way to access the function type information. This PR moves that concept over into `TypedFunction`. Much like `Typed`, this provides a way to access a function's `FunctionInfo`. By splitting this trait out, it helps to ensure the other traits are focused on a single responsibility. ### Internal Macros The original function PR (#13152) implemented `IntoFunction` using a macro which was passed into an `all_tuples!` macro invocation. Because we needed the same functionality for these new traits, this PR has copy+pasted that code for `ReflectFn`, `ReflectFnMut`, and `TypedFunction`— albeit with some differences between them. Originally, I was going to try and macro-ify the impls and where clauses such that we wouldn't have to straight up duplicate a lot of this logic. However, aside from being more complex in general, autocomplete just does not play nice with such heavily nested macros (tried in both RustRover and VSCode). And both of those problems told me that it just wasn't worth it: we need to ensure the crate is easily maintainable, even at the cost of duplicating code. So instead, I made sure to simplify the macro code by removing all fully-qualified syntax and cutting the where clauses down to the bare essentials, which helps to clean up a lot of the visual noise. I also tried my best to document the macro logic in certain areas (I may even add a bit more) to help with maintainability for future devs. ### Documentation Documentation for this module was a bit difficult for me. So many of these traits and types are very interconnected. And each trait/type has subtle differences that make documenting it in a single place, like at the module level, difficult to do cleanly. Describing the valid signatures is also challenging to do well. Hopefully what I have here is okay. I think I did an okay job, but let me know if there any thoughts on ways to improve it. We can also move such a task to a followup PR for more focused discussion. ## Testing You can test locally by running: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Added `DynamicClosure` struct - Added `DynamicClosureMut` struct - Added `IntoClosure` trait - Added `IntoClosureMut` trait - Added `ReflectFn` trait - Added `ReflectFnMut` trait - Added `TypedFunction` trait - `IntoFunction` now only works for standard Rust functions - `IntoFunction` no longer takes a lifetime parameter - `DynamicFunction::call` now only requires `&self` - Removed `DynamicFunction::call_once` - Changed the `IntoReturn::into_return` signature to include a where clause ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. ### `IntoClosure` `IntoFunction` now only works for standard Rust functions. Calling `IntoFunction::into_function` on a closure that captures references to its environment (either mutable or immutable), will no longer compile. Instead, you will need to use either `IntoClosure::into_closure` to create a `DynamicClosure` or `IntoClosureMut::into_closure_mut` to create a `DynamicClosureMut`, depending on your needs: ```rust let punct = String::from("!"); let print = |value: String| { println!("{value}{punct}"); }; // BEFORE let func: DynamicFunction = print.into_function(); // AFTER let func: DynamicClosure = print.into_closure(); ``` ### `IntoFunction` lifetime Additionally, `IntoFunction` no longer takes a lifetime parameter as it always expects a `'static` lifetime. Usages will need to remove any lifetime parameters: ```rust // BEFORE fn execute<'env, F: IntoFunction<'env, Marker>, Marker>(f: F) {/* ... */} // AFTER fn execute<F: IntoFunction<Marker>, Marker>(f: F) {/* ... */} ``` ### `IntoReturn` `IntoReturn::into_return` now has a where clause. Any manual implementors will need to add this where clause to their implementation.
2024-07-16 03:22:43 +00:00
let args = dbg!(ArgList::new().push_owned(2_i32));
let return_value = dbg!(function.call(args).unwrap());
reflect: implement the unique reflect rfc (#7207) # 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](https://github.com/bevyengine/bevy/pull/7207#discussion_r1083476050). - 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](https://github.com/bevyengine/rfcs/pull/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 <alice.i.cecile@gmail.com> Co-authored-by: Gino Valente <49806985+MrGVSV@users.noreply.github.com>
2024-08-12 17:01:41 +00:00
let value: Box<dyn PartialReflect> = return_value.unwrap_owned();
assert_eq!(value.try_take::<i32>().unwrap(), 5);
bevy_reflect: Add `DynamicClosure` and `DynamicClosureMut` (#14141) # Objective As mentioned in [this](https://github.com/bevyengine/bevy/pull/13152#issuecomment-2198387297) comment, creating a function registry (see #14098) is a bit difficult due to the requirements of `DynamicFunction`. Internally, a `DynamicFunction` contains a `Box<dyn FnMut>` (the function that reifies reflected arguments and calls the actual function), which requires `&mut self` in order to be called. This means that users would require a mutable reference to the function registry for it to be useful— which isn't great. And they can't clone the `DynamicFunction` either because cloning an `FnMut` isn't really feasible (wrapping it in an `Arc` would allow it to be cloned but we wouldn't be able to call the clone since we need a mutable reference to the `FnMut`, which we can't get with multiple `Arc`s still alive, requiring us to also slap in a `Mutex`, which adds additional overhead). And we don't want to just replace the `dyn FnMut` with `dyn Fn` as that would prevent reflecting closures that mutate their environment. Instead, we need to introduce a new type to split the requirements of `DynamicFunction`. ## Solution Introduce new types for representing closures. Specifically, this PR introduces `DynamicClosure` and `DynamicClosureMut`. Similar to how `IntoFunction` exists for `DynamicFunction`, two new traits were introduced: `IntoClosure` and `IntoClosureMut`. Now `DynamicFunction` stores a `dyn Fn` with a `'static` lifetime. `DynamicClosure` also uses a `dyn Fn` but has a lifetime, `'env`, tied to its environment. `DynamicClosureMut` is most like the old `DynamicFunction`, keeping the `dyn FnMut` and also typing its lifetime, `'env`, to the environment Here are some comparison tables: | | `DynamicFunction` | `DynamicClosure` | `DynamicClosureMut` | | - | ----------------- | ---------------- | ------------------- | | Callable with `&self` | ✅ | ✅ | ❌ | | Callable with `&mut self` | ✅ | ✅ | ✅ | | Allows for non-`'static` lifetimes | ❌ | ✅ | ✅ | | | `IntoFunction` | `IntoClosure` | `IntoClosureMut` | | - | -------------- | ------------- | ---------------- | | Convert `fn` functions | ✅ | ✅ | ✅ | | Convert `fn` methods | ✅ | ✅ | ✅ | | Convert anonymous functions | ✅ | ✅ | ✅ | | Convert closures that capture immutable references | ❌ | ✅ | ✅ | | Convert closures that capture mutable references | ❌ | ❌ | ✅ | | Convert closures that capture owned values | ❌[^1] | ✅ | ✅ | [^1]: Due to limitations in Rust, `IntoFunction` can't be implemented for just functions (unless we forced users to manually coerce them to function pointers first). So closures that meet the trait requirements _can technically_ be converted into a `DynamicFunction` as well. To both future-proof and reduce confusion, though, we'll just pretend like this isn't a thing. ```rust let mut list: Vec<i32> = vec![1, 2, 3]; // `replace` is a closure that captures a mutable reference to `list` let mut replace = |index: usize, value: i32| -> i32 { let old_value = list[index]; list[index] = value; old_value }; // Convert the closure into a dynamic closure using `IntoClosureMut::into_closure_mut` let mut func: DynamicClosureMut = replace.into_closure_mut(); // Dynamically call the closure: let args = ArgList::default().push_owned(1_usize).push_owned(-2_i32); let value = func.call_once(args).unwrap().unwrap_owned(); // Check the result: assert_eq!(value.take::<i32>().unwrap(), 2); assert_eq!(list, vec![1, -2, 3]); ``` ### `ReflectFn`/`ReflectFnMut` To make extending the function reflection system easier (the blanket impls for `IntoFunction`, `IntoClosure`, and `IntoClosureMut` are all incredibly short), this PR generalizes callables with two new traits: `ReflectFn` and `ReflectFnMut`. These traits mimic `Fn` and `FnMut` but allow for being called via reflection. In fact, their blanket implementations are identical save for `ReflectFn` being implemented over `Fn` types and `ReflectFnMut` being implemented over `FnMut` types. And just as `Fn` is a subtrait of `FnMut`, `ReflectFn` is a subtrait of `ReflectFnMut`. So anywhere that expects a `ReflectFnMut` can also be given a `ReflectFn`. To reiterate, these traits aren't 100% necessary. They were added in purely for extensibility. If we decide to split things up differently or add new traits/types in the future, then those changes should be much simpler to implement. ### `TypedFunction` Because of the split into `ReflectFn` and `ReflectFnMut`, we needed a new way to access the function type information. This PR moves that concept over into `TypedFunction`. Much like `Typed`, this provides a way to access a function's `FunctionInfo`. By splitting this trait out, it helps to ensure the other traits are focused on a single responsibility. ### Internal Macros The original function PR (#13152) implemented `IntoFunction` using a macro which was passed into an `all_tuples!` macro invocation. Because we needed the same functionality for these new traits, this PR has copy+pasted that code for `ReflectFn`, `ReflectFnMut`, and `TypedFunction`— albeit with some differences between them. Originally, I was going to try and macro-ify the impls and where clauses such that we wouldn't have to straight up duplicate a lot of this logic. However, aside from being more complex in general, autocomplete just does not play nice with such heavily nested macros (tried in both RustRover and VSCode). And both of those problems told me that it just wasn't worth it: we need to ensure the crate is easily maintainable, even at the cost of duplicating code. So instead, I made sure to simplify the macro code by removing all fully-qualified syntax and cutting the where clauses down to the bare essentials, which helps to clean up a lot of the visual noise. I also tried my best to document the macro logic in certain areas (I may even add a bit more) to help with maintainability for future devs. ### Documentation Documentation for this module was a bit difficult for me. So many of these traits and types are very interconnected. And each trait/type has subtle differences that make documenting it in a single place, like at the module level, difficult to do cleanly. Describing the valid signatures is also challenging to do well. Hopefully what I have here is okay. I think I did an okay job, but let me know if there any thoughts on ways to improve it. We can also move such a task to a followup PR for more focused discussion. ## Testing You can test locally by running: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Added `DynamicClosure` struct - Added `DynamicClosureMut` struct - Added `IntoClosure` trait - Added `IntoClosureMut` trait - Added `ReflectFn` trait - Added `ReflectFnMut` trait - Added `TypedFunction` trait - `IntoFunction` now only works for standard Rust functions - `IntoFunction` no longer takes a lifetime parameter - `DynamicFunction::call` now only requires `&self` - Removed `DynamicFunction::call_once` - Changed the `IntoReturn::into_return` signature to include a where clause ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. ### `IntoClosure` `IntoFunction` now only works for standard Rust functions. Calling `IntoFunction::into_function` on a closure that captures references to its environment (either mutable or immutable), will no longer compile. Instead, you will need to use either `IntoClosure::into_closure` to create a `DynamicClosure` or `IntoClosureMut::into_closure_mut` to create a `DynamicClosureMut`, depending on your needs: ```rust let punct = String::from("!"); let print = |value: String| { println!("{value}{punct}"); }; // BEFORE let func: DynamicFunction = print.into_function(); // AFTER let func: DynamicClosure = print.into_closure(); ``` ### `IntoFunction` lifetime Additionally, `IntoFunction` no longer takes a lifetime parameter as it always expects a `'static` lifetime. Usages will need to remove any lifetime parameters: ```rust // BEFORE fn execute<'env, F: IntoFunction<'env, Marker>, Marker>(f: F) {/* ... */} // AFTER fn execute<F: IntoFunction<Marker>, Marker>(f: F) {/* ... */} ``` ### `IntoReturn` `IntoReturn::into_return` now has a where clause. Any manual implementors will need to add this where clause to their implementation.
2024-07-16 03:22:43 +00:00
// We can also handle closures that capture their environment mutably
bevy_reflect: Function reflection terminology refactor (#14813) # Objective One of the changes in #14704 made `DynamicFunction` effectively the same as `DynamicClosure<'static>`. This change meant that the de facto function type would likely be `DynamicClosure<'static>` instead of the intended `DynamicFunction`, since the former is much more flexible. We _could_ explore ways of making `DynamicFunction` implement `Copy` using some unsafe code, but it likely wouldn't be worth it. And users would likely still reach for the convenience of `DynamicClosure<'static>` over the copy-ability of `DynamicFunction`. The goal of this PR is to fix this confusion between the two types. ## Solution Firstly, the `DynamicFunction` type was removed. Again, it was no different than `DynamicClosure<'static>` so it wasn't a huge deal to remove. Secondly, `DynamicClosure<'env>` and `DynamicClosureMut<'env>` were renamed to `DynamicFunction<'env>` and `DynamicFunctionMut<'env>`, respectively. Yes, we still ultimately kept the naming of `DynamicFunction`, but changed its behavior to that of `DynamicClosure<'env>`. We need a term to refer to both functions and closures, and "function" was the best option. [Originally](https://discord.com/channels/691052431525675048/1002362493634629796/1274091992162242710), I was going to go with "callable" as the replacement term to encompass both functions and closures (e.g. `DynamciCallable<'env>`). However, it was [suggested](https://discord.com/channels/691052431525675048/1002362493634629796/1274653581777047625) by @SkiFire13 that the simpler "function" term could be used instead. While "callable" is perhaps the better umbrella term—being truly ambiguous over functions and closures— "function" is more familiar, used more often, easier to discover, and is subjectively just "better-sounding". ## Testing Most changes are purely swapping type names or updating documentation, but you can verify everything still works by running the following command: ``` cargo test --package bevy_reflect ```
2024-08-19 21:52:36 +00:00
// using the `IntoFunctionMut` trait.
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
let mut count = 0;
bevy_reflect: Add `DynamicClosure` and `DynamicClosureMut` (#14141) # Objective As mentioned in [this](https://github.com/bevyengine/bevy/pull/13152#issuecomment-2198387297) comment, creating a function registry (see #14098) is a bit difficult due to the requirements of `DynamicFunction`. Internally, a `DynamicFunction` contains a `Box<dyn FnMut>` (the function that reifies reflected arguments and calls the actual function), which requires `&mut self` in order to be called. This means that users would require a mutable reference to the function registry for it to be useful— which isn't great. And they can't clone the `DynamicFunction` either because cloning an `FnMut` isn't really feasible (wrapping it in an `Arc` would allow it to be cloned but we wouldn't be able to call the clone since we need a mutable reference to the `FnMut`, which we can't get with multiple `Arc`s still alive, requiring us to also slap in a `Mutex`, which adds additional overhead). And we don't want to just replace the `dyn FnMut` with `dyn Fn` as that would prevent reflecting closures that mutate their environment. Instead, we need to introduce a new type to split the requirements of `DynamicFunction`. ## Solution Introduce new types for representing closures. Specifically, this PR introduces `DynamicClosure` and `DynamicClosureMut`. Similar to how `IntoFunction` exists for `DynamicFunction`, two new traits were introduced: `IntoClosure` and `IntoClosureMut`. Now `DynamicFunction` stores a `dyn Fn` with a `'static` lifetime. `DynamicClosure` also uses a `dyn Fn` but has a lifetime, `'env`, tied to its environment. `DynamicClosureMut` is most like the old `DynamicFunction`, keeping the `dyn FnMut` and also typing its lifetime, `'env`, to the environment Here are some comparison tables: | | `DynamicFunction` | `DynamicClosure` | `DynamicClosureMut` | | - | ----------------- | ---------------- | ------------------- | | Callable with `&self` | ✅ | ✅ | ❌ | | Callable with `&mut self` | ✅ | ✅ | ✅ | | Allows for non-`'static` lifetimes | ❌ | ✅ | ✅ | | | `IntoFunction` | `IntoClosure` | `IntoClosureMut` | | - | -------------- | ------------- | ---------------- | | Convert `fn` functions | ✅ | ✅ | ✅ | | Convert `fn` methods | ✅ | ✅ | ✅ | | Convert anonymous functions | ✅ | ✅ | ✅ | | Convert closures that capture immutable references | ❌ | ✅ | ✅ | | Convert closures that capture mutable references | ❌ | ❌ | ✅ | | Convert closures that capture owned values | ❌[^1] | ✅ | ✅ | [^1]: Due to limitations in Rust, `IntoFunction` can't be implemented for just functions (unless we forced users to manually coerce them to function pointers first). So closures that meet the trait requirements _can technically_ be converted into a `DynamicFunction` as well. To both future-proof and reduce confusion, though, we'll just pretend like this isn't a thing. ```rust let mut list: Vec<i32> = vec![1, 2, 3]; // `replace` is a closure that captures a mutable reference to `list` let mut replace = |index: usize, value: i32| -> i32 { let old_value = list[index]; list[index] = value; old_value }; // Convert the closure into a dynamic closure using `IntoClosureMut::into_closure_mut` let mut func: DynamicClosureMut = replace.into_closure_mut(); // Dynamically call the closure: let args = ArgList::default().push_owned(1_usize).push_owned(-2_i32); let value = func.call_once(args).unwrap().unwrap_owned(); // Check the result: assert_eq!(value.take::<i32>().unwrap(), 2); assert_eq!(list, vec![1, -2, 3]); ``` ### `ReflectFn`/`ReflectFnMut` To make extending the function reflection system easier (the blanket impls for `IntoFunction`, `IntoClosure`, and `IntoClosureMut` are all incredibly short), this PR generalizes callables with two new traits: `ReflectFn` and `ReflectFnMut`. These traits mimic `Fn` and `FnMut` but allow for being called via reflection. In fact, their blanket implementations are identical save for `ReflectFn` being implemented over `Fn` types and `ReflectFnMut` being implemented over `FnMut` types. And just as `Fn` is a subtrait of `FnMut`, `ReflectFn` is a subtrait of `ReflectFnMut`. So anywhere that expects a `ReflectFnMut` can also be given a `ReflectFn`. To reiterate, these traits aren't 100% necessary. They were added in purely for extensibility. If we decide to split things up differently or add new traits/types in the future, then those changes should be much simpler to implement. ### `TypedFunction` Because of the split into `ReflectFn` and `ReflectFnMut`, we needed a new way to access the function type information. This PR moves that concept over into `TypedFunction`. Much like `Typed`, this provides a way to access a function's `FunctionInfo`. By splitting this trait out, it helps to ensure the other traits are focused on a single responsibility. ### Internal Macros The original function PR (#13152) implemented `IntoFunction` using a macro which was passed into an `all_tuples!` macro invocation. Because we needed the same functionality for these new traits, this PR has copy+pasted that code for `ReflectFn`, `ReflectFnMut`, and `TypedFunction`— albeit with some differences between them. Originally, I was going to try and macro-ify the impls and where clauses such that we wouldn't have to straight up duplicate a lot of this logic. However, aside from being more complex in general, autocomplete just does not play nice with such heavily nested macros (tried in both RustRover and VSCode). And both of those problems told me that it just wasn't worth it: we need to ensure the crate is easily maintainable, even at the cost of duplicating code. So instead, I made sure to simplify the macro code by removing all fully-qualified syntax and cutting the where clauses down to the bare essentials, which helps to clean up a lot of the visual noise. I also tried my best to document the macro logic in certain areas (I may even add a bit more) to help with maintainability for future devs. ### Documentation Documentation for this module was a bit difficult for me. So many of these traits and types are very interconnected. And each trait/type has subtle differences that make documenting it in a single place, like at the module level, difficult to do cleanly. Describing the valid signatures is also challenging to do well. Hopefully what I have here is okay. I think I did an okay job, but let me know if there any thoughts on ways to improve it. We can also move such a task to a followup PR for more focused discussion. ## Testing You can test locally by running: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Added `DynamicClosure` struct - Added `DynamicClosureMut` struct - Added `IntoClosure` trait - Added `IntoClosureMut` trait - Added `ReflectFn` trait - Added `ReflectFnMut` trait - Added `TypedFunction` trait - `IntoFunction` now only works for standard Rust functions - `IntoFunction` no longer takes a lifetime parameter - `DynamicFunction::call` now only requires `&self` - Removed `DynamicFunction::call_once` - Changed the `IntoReturn::into_return` signature to include a where clause ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. ### `IntoClosure` `IntoFunction` now only works for standard Rust functions. Calling `IntoFunction::into_function` on a closure that captures references to its environment (either mutable or immutable), will no longer compile. Instead, you will need to use either `IntoClosure::into_closure` to create a `DynamicClosure` or `IntoClosureMut::into_closure_mut` to create a `DynamicClosureMut`, depending on your needs: ```rust let punct = String::from("!"); let print = |value: String| { println!("{value}{punct}"); }; // BEFORE let func: DynamicFunction = print.into_function(); // AFTER let func: DynamicClosure = print.into_closure(); ``` ### `IntoFunction` lifetime Additionally, `IntoFunction` no longer takes a lifetime parameter as it always expects a `'static` lifetime. Usages will need to remove any lifetime parameters: ```rust // BEFORE fn execute<'env, F: IntoFunction<'env, Marker>, Marker>(f: F) {/* ... */} // AFTER fn execute<F: IntoFunction<Marker>, Marker>(f: F) {/* ... */} ``` ### `IntoReturn` `IntoReturn::into_return` now has a where clause. Any manual implementors will need to add this where clause to their implementation.
2024-07-16 03:22:43 +00:00
let increment = |amount: i32| count += amount;
bevy_reflect: Function reflection terminology refactor (#14813) # Objective One of the changes in #14704 made `DynamicFunction` effectively the same as `DynamicClosure<'static>`. This change meant that the de facto function type would likely be `DynamicClosure<'static>` instead of the intended `DynamicFunction`, since the former is much more flexible. We _could_ explore ways of making `DynamicFunction` implement `Copy` using some unsafe code, but it likely wouldn't be worth it. And users would likely still reach for the convenience of `DynamicClosure<'static>` over the copy-ability of `DynamicFunction`. The goal of this PR is to fix this confusion between the two types. ## Solution Firstly, the `DynamicFunction` type was removed. Again, it was no different than `DynamicClosure<'static>` so it wasn't a huge deal to remove. Secondly, `DynamicClosure<'env>` and `DynamicClosureMut<'env>` were renamed to `DynamicFunction<'env>` and `DynamicFunctionMut<'env>`, respectively. Yes, we still ultimately kept the naming of `DynamicFunction`, but changed its behavior to that of `DynamicClosure<'env>`. We need a term to refer to both functions and closures, and "function" was the best option. [Originally](https://discord.com/channels/691052431525675048/1002362493634629796/1274091992162242710), I was going to go with "callable" as the replacement term to encompass both functions and closures (e.g. `DynamciCallable<'env>`). However, it was [suggested](https://discord.com/channels/691052431525675048/1002362493634629796/1274653581777047625) by @SkiFire13 that the simpler "function" term could be used instead. While "callable" is perhaps the better umbrella term—being truly ambiguous over functions and closures— "function" is more familiar, used more often, easier to discover, and is subjectively just "better-sounding". ## Testing Most changes are purely swapping type names or updating documentation, but you can verify everything still works by running the following command: ``` cargo test --package bevy_reflect ```
2024-08-19 21:52:36 +00:00
let closure: DynamicFunctionMut = dbg!(increment.into_function_mut());
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
let args = dbg!(ArgList::new().push_owned(5_i32));
bevy_reflect: Function reflection terminology refactor (#14813) # Objective One of the changes in #14704 made `DynamicFunction` effectively the same as `DynamicClosure<'static>`. This change meant that the de facto function type would likely be `DynamicClosure<'static>` instead of the intended `DynamicFunction`, since the former is much more flexible. We _could_ explore ways of making `DynamicFunction` implement `Copy` using some unsafe code, but it likely wouldn't be worth it. And users would likely still reach for the convenience of `DynamicClosure<'static>` over the copy-ability of `DynamicFunction`. The goal of this PR is to fix this confusion between the two types. ## Solution Firstly, the `DynamicFunction` type was removed. Again, it was no different than `DynamicClosure<'static>` so it wasn't a huge deal to remove. Secondly, `DynamicClosure<'env>` and `DynamicClosureMut<'env>` were renamed to `DynamicFunction<'env>` and `DynamicFunctionMut<'env>`, respectively. Yes, we still ultimately kept the naming of `DynamicFunction`, but changed its behavior to that of `DynamicClosure<'env>`. We need a term to refer to both functions and closures, and "function" was the best option. [Originally](https://discord.com/channels/691052431525675048/1002362493634629796/1274091992162242710), I was going to go with "callable" as the replacement term to encompass both functions and closures (e.g. `DynamciCallable<'env>`). However, it was [suggested](https://discord.com/channels/691052431525675048/1002362493634629796/1274653581777047625) by @SkiFire13 that the simpler "function" term could be used instead. While "callable" is perhaps the better umbrella term—being truly ambiguous over functions and closures— "function" is more familiar, used more often, easier to discover, and is subjectively just "better-sounding". ## Testing Most changes are purely swapping type names or updating documentation, but you can verify everything still works by running the following command: ``` cargo test --package bevy_reflect ```
2024-08-19 21:52:36 +00:00
// Because `DynamicFunctionMut` mutably borrows `total`,
bevy_reflect: Add `DynamicClosure` and `DynamicClosureMut` (#14141) # Objective As mentioned in [this](https://github.com/bevyengine/bevy/pull/13152#issuecomment-2198387297) comment, creating a function registry (see #14098) is a bit difficult due to the requirements of `DynamicFunction`. Internally, a `DynamicFunction` contains a `Box<dyn FnMut>` (the function that reifies reflected arguments and calls the actual function), which requires `&mut self` in order to be called. This means that users would require a mutable reference to the function registry for it to be useful— which isn't great. And they can't clone the `DynamicFunction` either because cloning an `FnMut` isn't really feasible (wrapping it in an `Arc` would allow it to be cloned but we wouldn't be able to call the clone since we need a mutable reference to the `FnMut`, which we can't get with multiple `Arc`s still alive, requiring us to also slap in a `Mutex`, which adds additional overhead). And we don't want to just replace the `dyn FnMut` with `dyn Fn` as that would prevent reflecting closures that mutate their environment. Instead, we need to introduce a new type to split the requirements of `DynamicFunction`. ## Solution Introduce new types for representing closures. Specifically, this PR introduces `DynamicClosure` and `DynamicClosureMut`. Similar to how `IntoFunction` exists for `DynamicFunction`, two new traits were introduced: `IntoClosure` and `IntoClosureMut`. Now `DynamicFunction` stores a `dyn Fn` with a `'static` lifetime. `DynamicClosure` also uses a `dyn Fn` but has a lifetime, `'env`, tied to its environment. `DynamicClosureMut` is most like the old `DynamicFunction`, keeping the `dyn FnMut` and also typing its lifetime, `'env`, to the environment Here are some comparison tables: | | `DynamicFunction` | `DynamicClosure` | `DynamicClosureMut` | | - | ----------------- | ---------------- | ------------------- | | Callable with `&self` | ✅ | ✅ | ❌ | | Callable with `&mut self` | ✅ | ✅ | ✅ | | Allows for non-`'static` lifetimes | ❌ | ✅ | ✅ | | | `IntoFunction` | `IntoClosure` | `IntoClosureMut` | | - | -------------- | ------------- | ---------------- | | Convert `fn` functions | ✅ | ✅ | ✅ | | Convert `fn` methods | ✅ | ✅ | ✅ | | Convert anonymous functions | ✅ | ✅ | ✅ | | Convert closures that capture immutable references | ❌ | ✅ | ✅ | | Convert closures that capture mutable references | ❌ | ❌ | ✅ | | Convert closures that capture owned values | ❌[^1] | ✅ | ✅ | [^1]: Due to limitations in Rust, `IntoFunction` can't be implemented for just functions (unless we forced users to manually coerce them to function pointers first). So closures that meet the trait requirements _can technically_ be converted into a `DynamicFunction` as well. To both future-proof and reduce confusion, though, we'll just pretend like this isn't a thing. ```rust let mut list: Vec<i32> = vec![1, 2, 3]; // `replace` is a closure that captures a mutable reference to `list` let mut replace = |index: usize, value: i32| -> i32 { let old_value = list[index]; list[index] = value; old_value }; // Convert the closure into a dynamic closure using `IntoClosureMut::into_closure_mut` let mut func: DynamicClosureMut = replace.into_closure_mut(); // Dynamically call the closure: let args = ArgList::default().push_owned(1_usize).push_owned(-2_i32); let value = func.call_once(args).unwrap().unwrap_owned(); // Check the result: assert_eq!(value.take::<i32>().unwrap(), 2); assert_eq!(list, vec![1, -2, 3]); ``` ### `ReflectFn`/`ReflectFnMut` To make extending the function reflection system easier (the blanket impls for `IntoFunction`, `IntoClosure`, and `IntoClosureMut` are all incredibly short), this PR generalizes callables with two new traits: `ReflectFn` and `ReflectFnMut`. These traits mimic `Fn` and `FnMut` but allow for being called via reflection. In fact, their blanket implementations are identical save for `ReflectFn` being implemented over `Fn` types and `ReflectFnMut` being implemented over `FnMut` types. And just as `Fn` is a subtrait of `FnMut`, `ReflectFn` is a subtrait of `ReflectFnMut`. So anywhere that expects a `ReflectFnMut` can also be given a `ReflectFn`. To reiterate, these traits aren't 100% necessary. They were added in purely for extensibility. If we decide to split things up differently or add new traits/types in the future, then those changes should be much simpler to implement. ### `TypedFunction` Because of the split into `ReflectFn` and `ReflectFnMut`, we needed a new way to access the function type information. This PR moves that concept over into `TypedFunction`. Much like `Typed`, this provides a way to access a function's `FunctionInfo`. By splitting this trait out, it helps to ensure the other traits are focused on a single responsibility. ### Internal Macros The original function PR (#13152) implemented `IntoFunction` using a macro which was passed into an `all_tuples!` macro invocation. Because we needed the same functionality for these new traits, this PR has copy+pasted that code for `ReflectFn`, `ReflectFnMut`, and `TypedFunction`— albeit with some differences between them. Originally, I was going to try and macro-ify the impls and where clauses such that we wouldn't have to straight up duplicate a lot of this logic. However, aside from being more complex in general, autocomplete just does not play nice with such heavily nested macros (tried in both RustRover and VSCode). And both of those problems told me that it just wasn't worth it: we need to ensure the crate is easily maintainable, even at the cost of duplicating code. So instead, I made sure to simplify the macro code by removing all fully-qualified syntax and cutting the where clauses down to the bare essentials, which helps to clean up a lot of the visual noise. I also tried my best to document the macro logic in certain areas (I may even add a bit more) to help with maintainability for future devs. ### Documentation Documentation for this module was a bit difficult for me. So many of these traits and types are very interconnected. And each trait/type has subtle differences that make documenting it in a single place, like at the module level, difficult to do cleanly. Describing the valid signatures is also challenging to do well. Hopefully what I have here is okay. I think I did an okay job, but let me know if there any thoughts on ways to improve it. We can also move such a task to a followup PR for more focused discussion. ## Testing You can test locally by running: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Added `DynamicClosure` struct - Added `DynamicClosureMut` struct - Added `IntoClosure` trait - Added `IntoClosureMut` trait - Added `ReflectFn` trait - Added `ReflectFnMut` trait - Added `TypedFunction` trait - `IntoFunction` now only works for standard Rust functions - `IntoFunction` no longer takes a lifetime parameter - `DynamicFunction::call` now only requires `&self` - Removed `DynamicFunction::call_once` - Changed the `IntoReturn::into_return` signature to include a where clause ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. ### `IntoClosure` `IntoFunction` now only works for standard Rust functions. Calling `IntoFunction::into_function` on a closure that captures references to its environment (either mutable or immutable), will no longer compile. Instead, you will need to use either `IntoClosure::into_closure` to create a `DynamicClosure` or `IntoClosureMut::into_closure_mut` to create a `DynamicClosureMut`, depending on your needs: ```rust let punct = String::from("!"); let print = |value: String| { println!("{value}{punct}"); }; // BEFORE let func: DynamicFunction = print.into_function(); // AFTER let func: DynamicClosure = print.into_closure(); ``` ### `IntoFunction` lifetime Additionally, `IntoFunction` no longer takes a lifetime parameter as it always expects a `'static` lifetime. Usages will need to remove any lifetime parameters: ```rust // BEFORE fn execute<'env, F: IntoFunction<'env, Marker>, Marker>(f: F) {/* ... */} // AFTER fn execute<F: IntoFunction<Marker>, Marker>(f: F) {/* ... */} ``` ### `IntoReturn` `IntoReturn::into_return` now has a where clause. Any manual implementors will need to add this where clause to their implementation.
2024-07-16 03:22:43 +00:00
// it will need to be dropped before `total` can be accessed again.
bevy_reflect: Function reflection terminology refactor (#14813) # Objective One of the changes in #14704 made `DynamicFunction` effectively the same as `DynamicClosure<'static>`. This change meant that the de facto function type would likely be `DynamicClosure<'static>` instead of the intended `DynamicFunction`, since the former is much more flexible. We _could_ explore ways of making `DynamicFunction` implement `Copy` using some unsafe code, but it likely wouldn't be worth it. And users would likely still reach for the convenience of `DynamicClosure<'static>` over the copy-ability of `DynamicFunction`. The goal of this PR is to fix this confusion between the two types. ## Solution Firstly, the `DynamicFunction` type was removed. Again, it was no different than `DynamicClosure<'static>` so it wasn't a huge deal to remove. Secondly, `DynamicClosure<'env>` and `DynamicClosureMut<'env>` were renamed to `DynamicFunction<'env>` and `DynamicFunctionMut<'env>`, respectively. Yes, we still ultimately kept the naming of `DynamicFunction`, but changed its behavior to that of `DynamicClosure<'env>`. We need a term to refer to both functions and closures, and "function" was the best option. [Originally](https://discord.com/channels/691052431525675048/1002362493634629796/1274091992162242710), I was going to go with "callable" as the replacement term to encompass both functions and closures (e.g. `DynamciCallable<'env>`). However, it was [suggested](https://discord.com/channels/691052431525675048/1002362493634629796/1274653581777047625) by @SkiFire13 that the simpler "function" term could be used instead. While "callable" is perhaps the better umbrella term—being truly ambiguous over functions and closures— "function" is more familiar, used more often, easier to discover, and is subjectively just "better-sounding". ## Testing Most changes are purely swapping type names or updating documentation, but you can verify everything still works by running the following command: ``` cargo test --package bevy_reflect ```
2024-08-19 21:52:36 +00:00
// This can be done manually with `drop(closure)` or by using the `DynamicFunctionMut::call_once` method.
bevy_reflect: Add `DynamicClosure` and `DynamicClosureMut` (#14141) # Objective As mentioned in [this](https://github.com/bevyengine/bevy/pull/13152#issuecomment-2198387297) comment, creating a function registry (see #14098) is a bit difficult due to the requirements of `DynamicFunction`. Internally, a `DynamicFunction` contains a `Box<dyn FnMut>` (the function that reifies reflected arguments and calls the actual function), which requires `&mut self` in order to be called. This means that users would require a mutable reference to the function registry for it to be useful— which isn't great. And they can't clone the `DynamicFunction` either because cloning an `FnMut` isn't really feasible (wrapping it in an `Arc` would allow it to be cloned but we wouldn't be able to call the clone since we need a mutable reference to the `FnMut`, which we can't get with multiple `Arc`s still alive, requiring us to also slap in a `Mutex`, which adds additional overhead). And we don't want to just replace the `dyn FnMut` with `dyn Fn` as that would prevent reflecting closures that mutate their environment. Instead, we need to introduce a new type to split the requirements of `DynamicFunction`. ## Solution Introduce new types for representing closures. Specifically, this PR introduces `DynamicClosure` and `DynamicClosureMut`. Similar to how `IntoFunction` exists for `DynamicFunction`, two new traits were introduced: `IntoClosure` and `IntoClosureMut`. Now `DynamicFunction` stores a `dyn Fn` with a `'static` lifetime. `DynamicClosure` also uses a `dyn Fn` but has a lifetime, `'env`, tied to its environment. `DynamicClosureMut` is most like the old `DynamicFunction`, keeping the `dyn FnMut` and also typing its lifetime, `'env`, to the environment Here are some comparison tables: | | `DynamicFunction` | `DynamicClosure` | `DynamicClosureMut` | | - | ----------------- | ---------------- | ------------------- | | Callable with `&self` | ✅ | ✅ | ❌ | | Callable with `&mut self` | ✅ | ✅ | ✅ | | Allows for non-`'static` lifetimes | ❌ | ✅ | ✅ | | | `IntoFunction` | `IntoClosure` | `IntoClosureMut` | | - | -------------- | ------------- | ---------------- | | Convert `fn` functions | ✅ | ✅ | ✅ | | Convert `fn` methods | ✅ | ✅ | ✅ | | Convert anonymous functions | ✅ | ✅ | ✅ | | Convert closures that capture immutable references | ❌ | ✅ | ✅ | | Convert closures that capture mutable references | ❌ | ❌ | ✅ | | Convert closures that capture owned values | ❌[^1] | ✅ | ✅ | [^1]: Due to limitations in Rust, `IntoFunction` can't be implemented for just functions (unless we forced users to manually coerce them to function pointers first). So closures that meet the trait requirements _can technically_ be converted into a `DynamicFunction` as well. To both future-proof and reduce confusion, though, we'll just pretend like this isn't a thing. ```rust let mut list: Vec<i32> = vec![1, 2, 3]; // `replace` is a closure that captures a mutable reference to `list` let mut replace = |index: usize, value: i32| -> i32 { let old_value = list[index]; list[index] = value; old_value }; // Convert the closure into a dynamic closure using `IntoClosureMut::into_closure_mut` let mut func: DynamicClosureMut = replace.into_closure_mut(); // Dynamically call the closure: let args = ArgList::default().push_owned(1_usize).push_owned(-2_i32); let value = func.call_once(args).unwrap().unwrap_owned(); // Check the result: assert_eq!(value.take::<i32>().unwrap(), 2); assert_eq!(list, vec![1, -2, 3]); ``` ### `ReflectFn`/`ReflectFnMut` To make extending the function reflection system easier (the blanket impls for `IntoFunction`, `IntoClosure`, and `IntoClosureMut` are all incredibly short), this PR generalizes callables with two new traits: `ReflectFn` and `ReflectFnMut`. These traits mimic `Fn` and `FnMut` but allow for being called via reflection. In fact, their blanket implementations are identical save for `ReflectFn` being implemented over `Fn` types and `ReflectFnMut` being implemented over `FnMut` types. And just as `Fn` is a subtrait of `FnMut`, `ReflectFn` is a subtrait of `ReflectFnMut`. So anywhere that expects a `ReflectFnMut` can also be given a `ReflectFn`. To reiterate, these traits aren't 100% necessary. They were added in purely for extensibility. If we decide to split things up differently or add new traits/types in the future, then those changes should be much simpler to implement. ### `TypedFunction` Because of the split into `ReflectFn` and `ReflectFnMut`, we needed a new way to access the function type information. This PR moves that concept over into `TypedFunction`. Much like `Typed`, this provides a way to access a function's `FunctionInfo`. By splitting this trait out, it helps to ensure the other traits are focused on a single responsibility. ### Internal Macros The original function PR (#13152) implemented `IntoFunction` using a macro which was passed into an `all_tuples!` macro invocation. Because we needed the same functionality for these new traits, this PR has copy+pasted that code for `ReflectFn`, `ReflectFnMut`, and `TypedFunction`— albeit with some differences between them. Originally, I was going to try and macro-ify the impls and where clauses such that we wouldn't have to straight up duplicate a lot of this logic. However, aside from being more complex in general, autocomplete just does not play nice with such heavily nested macros (tried in both RustRover and VSCode). And both of those problems told me that it just wasn't worth it: we need to ensure the crate is easily maintainable, even at the cost of duplicating code. So instead, I made sure to simplify the macro code by removing all fully-qualified syntax and cutting the where clauses down to the bare essentials, which helps to clean up a lot of the visual noise. I also tried my best to document the macro logic in certain areas (I may even add a bit more) to help with maintainability for future devs. ### Documentation Documentation for this module was a bit difficult for me. So many of these traits and types are very interconnected. And each trait/type has subtle differences that make documenting it in a single place, like at the module level, difficult to do cleanly. Describing the valid signatures is also challenging to do well. Hopefully what I have here is okay. I think I did an okay job, but let me know if there any thoughts on ways to improve it. We can also move such a task to a followup PR for more focused discussion. ## Testing You can test locally by running: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Added `DynamicClosure` struct - Added `DynamicClosureMut` struct - Added `IntoClosure` trait - Added `IntoClosureMut` trait - Added `ReflectFn` trait - Added `ReflectFnMut` trait - Added `TypedFunction` trait - `IntoFunction` now only works for standard Rust functions - `IntoFunction` no longer takes a lifetime parameter - `DynamicFunction::call` now only requires `&self` - Removed `DynamicFunction::call_once` - Changed the `IntoReturn::into_return` signature to include a where clause ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. ### `IntoClosure` `IntoFunction` now only works for standard Rust functions. Calling `IntoFunction::into_function` on a closure that captures references to its environment (either mutable or immutable), will no longer compile. Instead, you will need to use either `IntoClosure::into_closure` to create a `DynamicClosure` or `IntoClosureMut::into_closure_mut` to create a `DynamicClosureMut`, depending on your needs: ```rust let punct = String::from("!"); let print = |value: String| { println!("{value}{punct}"); }; // BEFORE let func: DynamicFunction = print.into_function(); // AFTER let func: DynamicClosure = print.into_closure(); ``` ### `IntoFunction` lifetime Additionally, `IntoFunction` no longer takes a lifetime parameter as it always expects a `'static` lifetime. Usages will need to remove any lifetime parameters: ```rust // BEFORE fn execute<'env, F: IntoFunction<'env, Marker>, Marker>(f: F) {/* ... */} // AFTER fn execute<F: IntoFunction<Marker>, Marker>(f: F) {/* ... */} ``` ### `IntoReturn` `IntoReturn::into_return` now has a where clause. Any manual implementors will need to add this where clause to their implementation.
2024-07-16 03:22:43 +00:00
dbg!(closure.call_once(args).unwrap());
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
assert_eq!(count, 5);
bevy_reflect: Function Overloading (Generic & Variadic Functions) (#15074) # Objective Currently function reflection requires users to manually monomorphize their generic functions. For example: ```rust fn add<T: Add<Output=T>>(a: T, b: T) -> T { a + b } // We have to specify the type of `T`: let reflect_add = add::<i32>.into_function(); ``` This PR doesn't aim to solve that problem—this is just a limitation in Rust. However, it also means that reflected functions can only ever work for a single monomorphization. If we wanted to support other types for `T`, we'd have to create a separate function for each one: ```rust let reflect_add_i32 = add::<i32>.into_function(); let reflect_add_u32 = add::<u32>.into_function(); let reflect_add_f32 = add::<f32>.into_function(); // ... ``` So in addition to requiring manual monomorphization, we also lose the benefit of having a single function handle multiple argument types. If a user wanted to create a small modding script that utilized function reflection, they'd have to either: - Store all sets of supported monomorphizations and require users to call the correct one - Write out some logic to find the correct function based on the given arguments While the first option would work, it wouldn't be very ergonomic. The second option is better, but it adds additional complexity to the user's logic—complexity that `bevy_reflect` could instead take on. ## Solution Introduce [function overloading](https://en.wikipedia.org/wiki/Function_overloading). A `DynamicFunction` can now be overloaded with other `DynamicFunction`s. We can rewrite the above code like so: ```rust let reflect_add = add::<i32> .into_function() .with_overload(add::<u32>) .with_overload(add::<f32>); ``` When invoked, the `DynamicFunction` will attempt to find a matching overload for the given set of arguments. And while I went into this PR only looking to improve generic function reflection, I accidentally added support for variadic functions as well (hence why I use the broader term "overload" over "generic"). ```rust // Supports 1 to 4 arguments let multiply_all = (|a: i32| a) .into_function() .with_overload(|a: i32, b: i32| a * b) .with_overload(|a: i32, b: i32, c: i32| a * b * c) .with_overload(|a: i32, b: i32, c: i32, d: i32| a * b * c * d); ``` This is simply an added bonus to this particular implementation. ~~Full variadic support (i.e. allowing for an indefinite number of arguments) will be added in a later PR.~~ I actually decided to limit the maximum number of arguments to 63 to supplement faster lookups, a reduced memory footprint, and faster cloning. ### Alternatives & Rationale I explored a few options for handling generic functions. This PR is the one I feel the most confident in, but I feel I should mention the others and why I ultimately didn't move forward with them. #### Adding `GenericDynamicFunction` **TL;DR:** Adding a distinct `GenericDynamicFunction` type unnecessarily splits and complicates the API. <details> <summary>Details</summary> My initial explorations involved a dedicated `GenericDynamicFunction` to contain and handle the mappings. This was initially started back when `DynamicFunction` was distinct from `DynamicClosure`. My goal was to not prevent us from being able to somehow make `DynamicFunction` implement `Copy`. But once we reverted back to a single `DynamicFunction`, that became a non-issue. But that aside, the real problem was that it created a split in the API. If I'm using a third-party library that uses function reflection, I have to know whether to request a `DynamicFunction` or a `GenericDynamicFunction`. I might not even know ahead of time which one I want. It might need to be determined at runtime. And if I'm creating a library, I might want a type to contain both `DynamicFunction` and `GenericDynamicFunction`. This might not be possible if, for example, I need to store the function in a `HashMap`. The other concern is with `IntoFunction`. Right now `DynamicFunction` trivially implements `IntoFunction` since it can just return itself. But what should `GenericDynamicFunction` do? It could return itself wrapped into a `DynamicFunction`, but then the API for `DynamicFunction` would have to account for this. So then what was the point of having a separate `GenericDynamicFunction` anyways? And even apart from `IntoFunction`, there's nothing stopping someone from manually creating a generic `DynamicFunction` through lying about its `FunctionInfo` and wrapping a `GenericDynamicFunction`. That being said, this is probably the "best" alternative if we added a `Function` trait and stored functions as `Box<dyn Function>`. However, I'm not convinced we gain much from this. Sure, we could keep the API for `DynamicFunction` the same, but consumers of `Function` will need to account for `GenericDynamicFunction` regardless (e.g. handling multiple `FunctionInfo`, a ranged argument count, etc.). And for all cases, except where using `DynamicFunction` directly, you end up treating them all like `GenericDynamicFunction`. Right now, if we did go with `GenericDynamicFunction`, the only major benefit we'd gain would be saving 24 bytes. If memory ever does become an issue here, we could swap over. But I think for the time being it's better for us to pursue a clearer mental model and end-user ergonomics through unification. </details> ##### Using the `FunctionRegistry` **TL;DR:** Having overloads only exist in the `FunctionRegistry` unnecessarily splits and complicates the API. <details> <summary>Details</summary> Another idea was to store the overloads in the `FunctionRegistry`. Users would then just call functions directly through the registry (i.e. `registry.call("my_func", my_args)`). I didn't go with this option because of how it specifically relies on the functions being registered. You'd not only always need access to the registry, but you'd need to ensure that the functions you want to call are even registered. It also means you can't just store a generic `DynamicFunction` on a type. Instead, you'll need to store the function's name and use that to look up the function in the registry—even if it's only ever used by that type. Doing so also removes all the benefits of `DynamicFunction`, such as the ability to pass it to functions accepting `IntoFunction`, modify it if needed, and so on. Like `GenericDynamicFunction` this introduces a split in the ecosystem: you either store `DynamicFunction`, store a string to look up the function, or force `DynamicFunction` to wrap your generic function anyways. Or worse yet: have `DynamicFunction` wrap the lookup function using `FunctionRegistryArc`. </details> #### Generic `ArgInfo` **TL;DR:** Allowing `ArgInfo` and `ReturnInfo` to store the generic information introduces a footgun when interpreting `FunctionInfo`. <details> <summary>Details</summary> Regardless of how we represent a generic function, one thing is clear: we need to be able to represent the information for such a function. This PR does so by introducing a `FunctionInfoType` enum to wrap one or more `FunctionInfo` values. Originally, I didn't do this. I had `ArgInfo` and `ReturnInfo` allow for generic types. This allowed us to have a single `FunctionInfo` to represent our function, but then I realized that it actually lies about our function. If we have two `ArgInfo` that both allow for either `i32` or `u32`, what does this tell us about our function? It turns out: nothing! We can't know whether our function takes `(i32, i32)`, `(u32, u32)`, `(i32, u32)`, or `(u32, i32)`. It therefore makes more sense to just represent a function with multiple `FunctionInfo` since that's really what it's made up of. </details> #### Flatten `FunctionInfo` **TL;DR:** Flattening removes additional per-overload information some users may desire and prevents us from adding more information in the future. <details> <summary>Details</summary> Why don't we just flatten multiple `FunctionInfo` into just one that can contain multiple signatures? This is something we could do, but I decided against it for a few reasons: - The only thing we'd be able to get rid of for each signature would be the `name`. While not enough to not do it, it doesn't really suggest we *have* to either. - Some consumers may want access to the names of the functions that make up the overloaded function. For example, to track a bug where an undesirable function is being added as an overload. Or to more easily locate the original function of an overload. - We may eventually allow for more information to be stored on `FunctionInfo`. For example, we may allow for documentation to be stored like we do for `TypeInfo`. Consumers of this documentation may want access to the documentation of each overload as they may provide documentation specific to that overload. </details> ## Testing This PR adds lots of tests and benchmarks, and also adds to the example. To run the tests: ``` cargo test --package bevy_reflect --all-features ``` To run the benchmarks: ``` cargo bench --bench reflect_function --all-features ``` To run the example: ``` cargo run --package bevy --example function_reflection --all-features ``` ### Benchmarks One of my goals with this PR was to leave the typical case of non-overloaded functions largely unaffected by the changes introduced in this PR. ~~And while the static size of `DynamicFunction` has increased by 17% (from 136 to 160 bytes), the performance has generally stayed the same~~ The static size of `DynamicFunction` has decreased from 136 to 112 bytes, while calling performance has generally stayed the same: | | `main` | 7d293ab | 252f3897d | |-------------------------------------|--------|---------|-----------| | `into/function` | 37 ns | 46 ns | 142 ns | | `with_overload/01_simple_overload` | - | 149 ns | 268 ns | | `with_overload/01_complex_overload` | - | 332 ns | 431 ns | | `with_overload/10_simple_overload` | - | 1266 ns | 2618 ns | | `with_overload/10_complex_overload` | - | 2544 ns | 4170 ns | | `call/function` | 57 ns | 58 ns | 61 ns | | `call/01_simple_overload` | - | 255 ns | 242 ns | | `call/01_complex_overload` | - | 595 ns | 431 ns | | `call/10_simple_overload` | - | 740 ns | 699 ns | | `call/10_complex_overload` | - | 1824 ns | 1618 ns | For the overloaded function tests, the leading number indicates how many overloads there are: `01` indicates 1 overload, `10` indicates 10 overloads. The `complex` cases have 10 unique generic types and 10 arguments, compared to the `simple` 1 generic type and 2 arguments. I aimed to prioritize the performance of calling the functions over creating them, hence creation speed tends to be a bit slower. There may be other optimizations we can look into but that's probably best saved for a future PR. The important bit is that the standard ~~`into/function`~~ and `call/function` benchmarks show minimal regressions. Since the latest changes, `into/function` does have some regressions, but again the priority was `call/function`. We can probably optimize `into/function` if needed in the future. --- ## Showcase Function reflection now supports [function overloading](https://en.wikipedia.org/wiki/Function_overloading)! This can be used to simulate generic functions: ```rust fn add<T: Add<Output=T>>(a: T, b: T) -> T { a + b } let reflect_add = add::<i32> .into_function() .with_overload(add::<u32>) .with_overload(add::<f32>); let args = ArgList::default().push_owned(25_i32).push_owned(75_i32); let result = func.call(args).unwrap().unwrap_owned(); assert_eq!(result.try_take::<i32>().unwrap(), 100); let args = ArgList::default().push_owned(25.0_f32).push_owned(75.0_f32); let result = func.call(args).unwrap().unwrap_owned(); assert_eq!(result.try_take::<f32>().unwrap(), 100.0); ``` You can also simulate variadic functions: ```rust #[derive(Reflect, PartialEq, Debug)] struct Player { name: Option<String>, health: u32, } // Creates a `Player` with one of the following: // - No name and 100 health // - A name and 100 health // - No name and custom health // - A name and custom health let create_player = (|| Player { name: None, health: 100, }) .into_function() .with_overload(|name: String| Player { name: Some(name), health: 100, }) .with_overload(|health: u32| Player { name: None, health }) .with_overload(|name: String, health: u32| Player { name: Some(name), health, }); let args = ArgList::default() .push_owned(String::from("Urist")) .push_owned(55_u32); let player = create_player .call(args) .unwrap() .unwrap_owned() .try_take::<Player>() .unwrap(); assert_eq!( player, Player { name: Some(String::from("Urist")), health: 55 } ); ```
2024-12-10 01:51:47 +00:00
// Generic functions can also be converted into a `DynamicFunction`,
// however, they will need to be manually monomorphized first.
fn stringify<T: ToString>(value: T) -> String {
value.to_string()
}
// We have to manually specify the concrete generic type we want to use.
let function = stringify::<i32>.into_function();
let args = ArgList::new().push_owned(123_i32);
let return_value = function.call(args).unwrap();
let value: Box<dyn PartialReflect> = return_value.unwrap_owned();
assert_eq!(value.try_take::<String>().unwrap(), "123");
// To make things a little easier, we can also "overload" functions.
// This makes it so that a single `DynamicFunction` can represent multiple functions,
// and the correct one is chosen based on the types of the arguments.
// Each function overload must have a unique argument signature.
let function = stringify::<i32>
.into_function()
.with_overload(stringify::<f32>);
// Now our `function` accepts both `i32` and `f32` arguments.
let args = ArgList::new().push_owned(1.23_f32);
let return_value = function.call(args).unwrap();
let value: Box<dyn PartialReflect> = return_value.unwrap_owned();
assert_eq!(value.try_take::<String>().unwrap(), "1.23");
// Function overloading even allows us to have a variable number of arguments.
let function = (|| 0)
.into_function()
.with_overload(|a: i32| a)
.with_overload(|a: i32, b: i32| a + b)
.with_overload(|a: i32, b: i32, c: i32| a + b + c);
let args = ArgList::new()
.push_owned(1_i32)
.push_owned(2_i32)
.push_owned(3_i32);
let return_value = function.call(args).unwrap();
let value: Box<dyn PartialReflect> = return_value.unwrap_owned();
assert_eq!(value.try_take::<i32>().unwrap(), 6);
// As stated earlier, `IntoFunction` works for many kinds of simple functions.
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
// Functions with non-reflectable arguments or return values may not be able to be converted.
bevy_reflect: Add `DynamicClosure` and `DynamicClosureMut` (#14141) # Objective As mentioned in [this](https://github.com/bevyengine/bevy/pull/13152#issuecomment-2198387297) comment, creating a function registry (see #14098) is a bit difficult due to the requirements of `DynamicFunction`. Internally, a `DynamicFunction` contains a `Box<dyn FnMut>` (the function that reifies reflected arguments and calls the actual function), which requires `&mut self` in order to be called. This means that users would require a mutable reference to the function registry for it to be useful— which isn't great. And they can't clone the `DynamicFunction` either because cloning an `FnMut` isn't really feasible (wrapping it in an `Arc` would allow it to be cloned but we wouldn't be able to call the clone since we need a mutable reference to the `FnMut`, which we can't get with multiple `Arc`s still alive, requiring us to also slap in a `Mutex`, which adds additional overhead). And we don't want to just replace the `dyn FnMut` with `dyn Fn` as that would prevent reflecting closures that mutate their environment. Instead, we need to introduce a new type to split the requirements of `DynamicFunction`. ## Solution Introduce new types for representing closures. Specifically, this PR introduces `DynamicClosure` and `DynamicClosureMut`. Similar to how `IntoFunction` exists for `DynamicFunction`, two new traits were introduced: `IntoClosure` and `IntoClosureMut`. Now `DynamicFunction` stores a `dyn Fn` with a `'static` lifetime. `DynamicClosure` also uses a `dyn Fn` but has a lifetime, `'env`, tied to its environment. `DynamicClosureMut` is most like the old `DynamicFunction`, keeping the `dyn FnMut` and also typing its lifetime, `'env`, to the environment Here are some comparison tables: | | `DynamicFunction` | `DynamicClosure` | `DynamicClosureMut` | | - | ----------------- | ---------------- | ------------------- | | Callable with `&self` | ✅ | ✅ | ❌ | | Callable with `&mut self` | ✅ | ✅ | ✅ | | Allows for non-`'static` lifetimes | ❌ | ✅ | ✅ | | | `IntoFunction` | `IntoClosure` | `IntoClosureMut` | | - | -------------- | ------------- | ---------------- | | Convert `fn` functions | ✅ | ✅ | ✅ | | Convert `fn` methods | ✅ | ✅ | ✅ | | Convert anonymous functions | ✅ | ✅ | ✅ | | Convert closures that capture immutable references | ❌ | ✅ | ✅ | | Convert closures that capture mutable references | ❌ | ❌ | ✅ | | Convert closures that capture owned values | ❌[^1] | ✅ | ✅ | [^1]: Due to limitations in Rust, `IntoFunction` can't be implemented for just functions (unless we forced users to manually coerce them to function pointers first). So closures that meet the trait requirements _can technically_ be converted into a `DynamicFunction` as well. To both future-proof and reduce confusion, though, we'll just pretend like this isn't a thing. ```rust let mut list: Vec<i32> = vec![1, 2, 3]; // `replace` is a closure that captures a mutable reference to `list` let mut replace = |index: usize, value: i32| -> i32 { let old_value = list[index]; list[index] = value; old_value }; // Convert the closure into a dynamic closure using `IntoClosureMut::into_closure_mut` let mut func: DynamicClosureMut = replace.into_closure_mut(); // Dynamically call the closure: let args = ArgList::default().push_owned(1_usize).push_owned(-2_i32); let value = func.call_once(args).unwrap().unwrap_owned(); // Check the result: assert_eq!(value.take::<i32>().unwrap(), 2); assert_eq!(list, vec![1, -2, 3]); ``` ### `ReflectFn`/`ReflectFnMut` To make extending the function reflection system easier (the blanket impls for `IntoFunction`, `IntoClosure`, and `IntoClosureMut` are all incredibly short), this PR generalizes callables with two new traits: `ReflectFn` and `ReflectFnMut`. These traits mimic `Fn` and `FnMut` but allow for being called via reflection. In fact, their blanket implementations are identical save for `ReflectFn` being implemented over `Fn` types and `ReflectFnMut` being implemented over `FnMut` types. And just as `Fn` is a subtrait of `FnMut`, `ReflectFn` is a subtrait of `ReflectFnMut`. So anywhere that expects a `ReflectFnMut` can also be given a `ReflectFn`. To reiterate, these traits aren't 100% necessary. They were added in purely for extensibility. If we decide to split things up differently or add new traits/types in the future, then those changes should be much simpler to implement. ### `TypedFunction` Because of the split into `ReflectFn` and `ReflectFnMut`, we needed a new way to access the function type information. This PR moves that concept over into `TypedFunction`. Much like `Typed`, this provides a way to access a function's `FunctionInfo`. By splitting this trait out, it helps to ensure the other traits are focused on a single responsibility. ### Internal Macros The original function PR (#13152) implemented `IntoFunction` using a macro which was passed into an `all_tuples!` macro invocation. Because we needed the same functionality for these new traits, this PR has copy+pasted that code for `ReflectFn`, `ReflectFnMut`, and `TypedFunction`— albeit with some differences between them. Originally, I was going to try and macro-ify the impls and where clauses such that we wouldn't have to straight up duplicate a lot of this logic. However, aside from being more complex in general, autocomplete just does not play nice with such heavily nested macros (tried in both RustRover and VSCode). And both of those problems told me that it just wasn't worth it: we need to ensure the crate is easily maintainable, even at the cost of duplicating code. So instead, I made sure to simplify the macro code by removing all fully-qualified syntax and cutting the where clauses down to the bare essentials, which helps to clean up a lot of the visual noise. I also tried my best to document the macro logic in certain areas (I may even add a bit more) to help with maintainability for future devs. ### Documentation Documentation for this module was a bit difficult for me. So many of these traits and types are very interconnected. And each trait/type has subtle differences that make documenting it in a single place, like at the module level, difficult to do cleanly. Describing the valid signatures is also challenging to do well. Hopefully what I have here is okay. I think I did an okay job, but let me know if there any thoughts on ways to improve it. We can also move such a task to a followup PR for more focused discussion. ## Testing You can test locally by running: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Added `DynamicClosure` struct - Added `DynamicClosureMut` struct - Added `IntoClosure` trait - Added `IntoClosureMut` trait - Added `ReflectFn` trait - Added `ReflectFnMut` trait - Added `TypedFunction` trait - `IntoFunction` now only works for standard Rust functions - `IntoFunction` no longer takes a lifetime parameter - `DynamicFunction::call` now only requires `&self` - Removed `DynamicFunction::call_once` - Changed the `IntoReturn::into_return` signature to include a where clause ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. ### `IntoClosure` `IntoFunction` now only works for standard Rust functions. Calling `IntoFunction::into_function` on a closure that captures references to its environment (either mutable or immutable), will no longer compile. Instead, you will need to use either `IntoClosure::into_closure` to create a `DynamicClosure` or `IntoClosureMut::into_closure_mut` to create a `DynamicClosureMut`, depending on your needs: ```rust let punct = String::from("!"); let print = |value: String| { println!("{value}{punct}"); }; // BEFORE let func: DynamicFunction = print.into_function(); // AFTER let func: DynamicClosure = print.into_closure(); ``` ### `IntoFunction` lifetime Additionally, `IntoFunction` no longer takes a lifetime parameter as it always expects a `'static` lifetime. Usages will need to remove any lifetime parameters: ```rust // BEFORE fn execute<'env, F: IntoFunction<'env, Marker>, Marker>(f: F) {/* ... */} // AFTER fn execute<F: IntoFunction<Marker>, Marker>(f: F) {/* ... */} ``` ### `IntoReturn` `IntoReturn::into_return` now has a where clause. Any manual implementors will need to add this where clause to their implementation.
2024-07-16 03:22:43 +00:00
// Generic functions are also not supported (unless manually monomorphized like `foo::<i32>.into_function()`).
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
// Additionally, the lifetime of the return value is tied to the lifetime of the first argument.
// However, this means that many methods (i.e. functions with a `self` parameter) are also supported:
#[derive(Reflect, Default)]
struct Data {
value: String,
}
impl Data {
fn set_value(&mut self, value: String) {
self.value = value;
}
// Note that only `&'static str` implements `Reflect`.
// To get around this limitation we can use `&String` instead.
fn get_value(&self) -> &String {
&self.value
}
}
let mut data = Data::default();
bevy_reflect: Add `DynamicClosure` and `DynamicClosureMut` (#14141) # Objective As mentioned in [this](https://github.com/bevyengine/bevy/pull/13152#issuecomment-2198387297) comment, creating a function registry (see #14098) is a bit difficult due to the requirements of `DynamicFunction`. Internally, a `DynamicFunction` contains a `Box<dyn FnMut>` (the function that reifies reflected arguments and calls the actual function), which requires `&mut self` in order to be called. This means that users would require a mutable reference to the function registry for it to be useful— which isn't great. And they can't clone the `DynamicFunction` either because cloning an `FnMut` isn't really feasible (wrapping it in an `Arc` would allow it to be cloned but we wouldn't be able to call the clone since we need a mutable reference to the `FnMut`, which we can't get with multiple `Arc`s still alive, requiring us to also slap in a `Mutex`, which adds additional overhead). And we don't want to just replace the `dyn FnMut` with `dyn Fn` as that would prevent reflecting closures that mutate their environment. Instead, we need to introduce a new type to split the requirements of `DynamicFunction`. ## Solution Introduce new types for representing closures. Specifically, this PR introduces `DynamicClosure` and `DynamicClosureMut`. Similar to how `IntoFunction` exists for `DynamicFunction`, two new traits were introduced: `IntoClosure` and `IntoClosureMut`. Now `DynamicFunction` stores a `dyn Fn` with a `'static` lifetime. `DynamicClosure` also uses a `dyn Fn` but has a lifetime, `'env`, tied to its environment. `DynamicClosureMut` is most like the old `DynamicFunction`, keeping the `dyn FnMut` and also typing its lifetime, `'env`, to the environment Here are some comparison tables: | | `DynamicFunction` | `DynamicClosure` | `DynamicClosureMut` | | - | ----------------- | ---------------- | ------------------- | | Callable with `&self` | ✅ | ✅ | ❌ | | Callable with `&mut self` | ✅ | ✅ | ✅ | | Allows for non-`'static` lifetimes | ❌ | ✅ | ✅ | | | `IntoFunction` | `IntoClosure` | `IntoClosureMut` | | - | -------------- | ------------- | ---------------- | | Convert `fn` functions | ✅ | ✅ | ✅ | | Convert `fn` methods | ✅ | ✅ | ✅ | | Convert anonymous functions | ✅ | ✅ | ✅ | | Convert closures that capture immutable references | ❌ | ✅ | ✅ | | Convert closures that capture mutable references | ❌ | ❌ | ✅ | | Convert closures that capture owned values | ❌[^1] | ✅ | ✅ | [^1]: Due to limitations in Rust, `IntoFunction` can't be implemented for just functions (unless we forced users to manually coerce them to function pointers first). So closures that meet the trait requirements _can technically_ be converted into a `DynamicFunction` as well. To both future-proof and reduce confusion, though, we'll just pretend like this isn't a thing. ```rust let mut list: Vec<i32> = vec![1, 2, 3]; // `replace` is a closure that captures a mutable reference to `list` let mut replace = |index: usize, value: i32| -> i32 { let old_value = list[index]; list[index] = value; old_value }; // Convert the closure into a dynamic closure using `IntoClosureMut::into_closure_mut` let mut func: DynamicClosureMut = replace.into_closure_mut(); // Dynamically call the closure: let args = ArgList::default().push_owned(1_usize).push_owned(-2_i32); let value = func.call_once(args).unwrap().unwrap_owned(); // Check the result: assert_eq!(value.take::<i32>().unwrap(), 2); assert_eq!(list, vec![1, -2, 3]); ``` ### `ReflectFn`/`ReflectFnMut` To make extending the function reflection system easier (the blanket impls for `IntoFunction`, `IntoClosure`, and `IntoClosureMut` are all incredibly short), this PR generalizes callables with two new traits: `ReflectFn` and `ReflectFnMut`. These traits mimic `Fn` and `FnMut` but allow for being called via reflection. In fact, their blanket implementations are identical save for `ReflectFn` being implemented over `Fn` types and `ReflectFnMut` being implemented over `FnMut` types. And just as `Fn` is a subtrait of `FnMut`, `ReflectFn` is a subtrait of `ReflectFnMut`. So anywhere that expects a `ReflectFnMut` can also be given a `ReflectFn`. To reiterate, these traits aren't 100% necessary. They were added in purely for extensibility. If we decide to split things up differently or add new traits/types in the future, then those changes should be much simpler to implement. ### `TypedFunction` Because of the split into `ReflectFn` and `ReflectFnMut`, we needed a new way to access the function type information. This PR moves that concept over into `TypedFunction`. Much like `Typed`, this provides a way to access a function's `FunctionInfo`. By splitting this trait out, it helps to ensure the other traits are focused on a single responsibility. ### Internal Macros The original function PR (#13152) implemented `IntoFunction` using a macro which was passed into an `all_tuples!` macro invocation. Because we needed the same functionality for these new traits, this PR has copy+pasted that code for `ReflectFn`, `ReflectFnMut`, and `TypedFunction`— albeit with some differences between them. Originally, I was going to try and macro-ify the impls and where clauses such that we wouldn't have to straight up duplicate a lot of this logic. However, aside from being more complex in general, autocomplete just does not play nice with such heavily nested macros (tried in both RustRover and VSCode). And both of those problems told me that it just wasn't worth it: we need to ensure the crate is easily maintainable, even at the cost of duplicating code. So instead, I made sure to simplify the macro code by removing all fully-qualified syntax and cutting the where clauses down to the bare essentials, which helps to clean up a lot of the visual noise. I also tried my best to document the macro logic in certain areas (I may even add a bit more) to help with maintainability for future devs. ### Documentation Documentation for this module was a bit difficult for me. So many of these traits and types are very interconnected. And each trait/type has subtle differences that make documenting it in a single place, like at the module level, difficult to do cleanly. Describing the valid signatures is also challenging to do well. Hopefully what I have here is okay. I think I did an okay job, but let me know if there any thoughts on ways to improve it. We can also move such a task to a followup PR for more focused discussion. ## Testing You can test locally by running: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Added `DynamicClosure` struct - Added `DynamicClosureMut` struct - Added `IntoClosure` trait - Added `IntoClosureMut` trait - Added `ReflectFn` trait - Added `ReflectFnMut` trait - Added `TypedFunction` trait - `IntoFunction` now only works for standard Rust functions - `IntoFunction` no longer takes a lifetime parameter - `DynamicFunction::call` now only requires `&self` - Removed `DynamicFunction::call_once` - Changed the `IntoReturn::into_return` signature to include a where clause ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. ### `IntoClosure` `IntoFunction` now only works for standard Rust functions. Calling `IntoFunction::into_function` on a closure that captures references to its environment (either mutable or immutable), will no longer compile. Instead, you will need to use either `IntoClosure::into_closure` to create a `DynamicClosure` or `IntoClosureMut::into_closure_mut` to create a `DynamicClosureMut`, depending on your needs: ```rust let punct = String::from("!"); let print = |value: String| { println!("{value}{punct}"); }; // BEFORE let func: DynamicFunction = print.into_function(); // AFTER let func: DynamicClosure = print.into_closure(); ``` ### `IntoFunction` lifetime Additionally, `IntoFunction` no longer takes a lifetime parameter as it always expects a `'static` lifetime. Usages will need to remove any lifetime parameters: ```rust // BEFORE fn execute<'env, F: IntoFunction<'env, Marker>, Marker>(f: F) {/* ... */} // AFTER fn execute<F: IntoFunction<Marker>, Marker>(f: F) {/* ... */} ``` ### `IntoReturn` `IntoReturn::into_return` now has a where clause. Any manual implementors will need to add this where clause to their implementation.
2024-07-16 03:22:43 +00:00
let set_value = dbg!(Data::set_value.into_function());
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
let args = dbg!(ArgList::new().push_mut(&mut data)).push_owned(String::from("Hello, world!"));
dbg!(set_value.call(args).unwrap());
assert_eq!(data.value, "Hello, world!");
bevy_reflect: Add `DynamicClosure` and `DynamicClosureMut` (#14141) # Objective As mentioned in [this](https://github.com/bevyengine/bevy/pull/13152#issuecomment-2198387297) comment, creating a function registry (see #14098) is a bit difficult due to the requirements of `DynamicFunction`. Internally, a `DynamicFunction` contains a `Box<dyn FnMut>` (the function that reifies reflected arguments and calls the actual function), which requires `&mut self` in order to be called. This means that users would require a mutable reference to the function registry for it to be useful— which isn't great. And they can't clone the `DynamicFunction` either because cloning an `FnMut` isn't really feasible (wrapping it in an `Arc` would allow it to be cloned but we wouldn't be able to call the clone since we need a mutable reference to the `FnMut`, which we can't get with multiple `Arc`s still alive, requiring us to also slap in a `Mutex`, which adds additional overhead). And we don't want to just replace the `dyn FnMut` with `dyn Fn` as that would prevent reflecting closures that mutate their environment. Instead, we need to introduce a new type to split the requirements of `DynamicFunction`. ## Solution Introduce new types for representing closures. Specifically, this PR introduces `DynamicClosure` and `DynamicClosureMut`. Similar to how `IntoFunction` exists for `DynamicFunction`, two new traits were introduced: `IntoClosure` and `IntoClosureMut`. Now `DynamicFunction` stores a `dyn Fn` with a `'static` lifetime. `DynamicClosure` also uses a `dyn Fn` but has a lifetime, `'env`, tied to its environment. `DynamicClosureMut` is most like the old `DynamicFunction`, keeping the `dyn FnMut` and also typing its lifetime, `'env`, to the environment Here are some comparison tables: | | `DynamicFunction` | `DynamicClosure` | `DynamicClosureMut` | | - | ----------------- | ---------------- | ------------------- | | Callable with `&self` | ✅ | ✅ | ❌ | | Callable with `&mut self` | ✅ | ✅ | ✅ | | Allows for non-`'static` lifetimes | ❌ | ✅ | ✅ | | | `IntoFunction` | `IntoClosure` | `IntoClosureMut` | | - | -------------- | ------------- | ---------------- | | Convert `fn` functions | ✅ | ✅ | ✅ | | Convert `fn` methods | ✅ | ✅ | ✅ | | Convert anonymous functions | ✅ | ✅ | ✅ | | Convert closures that capture immutable references | ❌ | ✅ | ✅ | | Convert closures that capture mutable references | ❌ | ❌ | ✅ | | Convert closures that capture owned values | ❌[^1] | ✅ | ✅ | [^1]: Due to limitations in Rust, `IntoFunction` can't be implemented for just functions (unless we forced users to manually coerce them to function pointers first). So closures that meet the trait requirements _can technically_ be converted into a `DynamicFunction` as well. To both future-proof and reduce confusion, though, we'll just pretend like this isn't a thing. ```rust let mut list: Vec<i32> = vec![1, 2, 3]; // `replace` is a closure that captures a mutable reference to `list` let mut replace = |index: usize, value: i32| -> i32 { let old_value = list[index]; list[index] = value; old_value }; // Convert the closure into a dynamic closure using `IntoClosureMut::into_closure_mut` let mut func: DynamicClosureMut = replace.into_closure_mut(); // Dynamically call the closure: let args = ArgList::default().push_owned(1_usize).push_owned(-2_i32); let value = func.call_once(args).unwrap().unwrap_owned(); // Check the result: assert_eq!(value.take::<i32>().unwrap(), 2); assert_eq!(list, vec![1, -2, 3]); ``` ### `ReflectFn`/`ReflectFnMut` To make extending the function reflection system easier (the blanket impls for `IntoFunction`, `IntoClosure`, and `IntoClosureMut` are all incredibly short), this PR generalizes callables with two new traits: `ReflectFn` and `ReflectFnMut`. These traits mimic `Fn` and `FnMut` but allow for being called via reflection. In fact, their blanket implementations are identical save for `ReflectFn` being implemented over `Fn` types and `ReflectFnMut` being implemented over `FnMut` types. And just as `Fn` is a subtrait of `FnMut`, `ReflectFn` is a subtrait of `ReflectFnMut`. So anywhere that expects a `ReflectFnMut` can also be given a `ReflectFn`. To reiterate, these traits aren't 100% necessary. They were added in purely for extensibility. If we decide to split things up differently or add new traits/types in the future, then those changes should be much simpler to implement. ### `TypedFunction` Because of the split into `ReflectFn` and `ReflectFnMut`, we needed a new way to access the function type information. This PR moves that concept over into `TypedFunction`. Much like `Typed`, this provides a way to access a function's `FunctionInfo`. By splitting this trait out, it helps to ensure the other traits are focused on a single responsibility. ### Internal Macros The original function PR (#13152) implemented `IntoFunction` using a macro which was passed into an `all_tuples!` macro invocation. Because we needed the same functionality for these new traits, this PR has copy+pasted that code for `ReflectFn`, `ReflectFnMut`, and `TypedFunction`— albeit with some differences between them. Originally, I was going to try and macro-ify the impls and where clauses such that we wouldn't have to straight up duplicate a lot of this logic. However, aside from being more complex in general, autocomplete just does not play nice with such heavily nested macros (tried in both RustRover and VSCode). And both of those problems told me that it just wasn't worth it: we need to ensure the crate is easily maintainable, even at the cost of duplicating code. So instead, I made sure to simplify the macro code by removing all fully-qualified syntax and cutting the where clauses down to the bare essentials, which helps to clean up a lot of the visual noise. I also tried my best to document the macro logic in certain areas (I may even add a bit more) to help with maintainability for future devs. ### Documentation Documentation for this module was a bit difficult for me. So many of these traits and types are very interconnected. And each trait/type has subtle differences that make documenting it in a single place, like at the module level, difficult to do cleanly. Describing the valid signatures is also challenging to do well. Hopefully what I have here is okay. I think I did an okay job, but let me know if there any thoughts on ways to improve it. We can also move such a task to a followup PR for more focused discussion. ## Testing You can test locally by running: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Added `DynamicClosure` struct - Added `DynamicClosureMut` struct - Added `IntoClosure` trait - Added `IntoClosureMut` trait - Added `ReflectFn` trait - Added `ReflectFnMut` trait - Added `TypedFunction` trait - `IntoFunction` now only works for standard Rust functions - `IntoFunction` no longer takes a lifetime parameter - `DynamicFunction::call` now only requires `&self` - Removed `DynamicFunction::call_once` - Changed the `IntoReturn::into_return` signature to include a where clause ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. ### `IntoClosure` `IntoFunction` now only works for standard Rust functions. Calling `IntoFunction::into_function` on a closure that captures references to its environment (either mutable or immutable), will no longer compile. Instead, you will need to use either `IntoClosure::into_closure` to create a `DynamicClosure` or `IntoClosureMut::into_closure_mut` to create a `DynamicClosureMut`, depending on your needs: ```rust let punct = String::from("!"); let print = |value: String| { println!("{value}{punct}"); }; // BEFORE let func: DynamicFunction = print.into_function(); // AFTER let func: DynamicClosure = print.into_closure(); ``` ### `IntoFunction` lifetime Additionally, `IntoFunction` no longer takes a lifetime parameter as it always expects a `'static` lifetime. Usages will need to remove any lifetime parameters: ```rust // BEFORE fn execute<'env, F: IntoFunction<'env, Marker>, Marker>(f: F) {/* ... */} // AFTER fn execute<F: IntoFunction<Marker>, Marker>(f: F) {/* ... */} ``` ### `IntoReturn` `IntoReturn::into_return` now has a where clause. Any manual implementors will need to add this where clause to their implementation.
2024-07-16 03:22:43 +00:00
let get_value = dbg!(Data::get_value.into_function());
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
let args = dbg!(ArgList::new().push_ref(&data));
let return_value = dbg!(get_value.call(args).unwrap());
reflect: implement the unique reflect rfc (#7207) # 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](https://github.com/bevyengine/bevy/pull/7207#discussion_r1083476050). - 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](https://github.com/bevyengine/rfcs/pull/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 <alice.i.cecile@gmail.com> Co-authored-by: Gino Valente <49806985+MrGVSV@users.noreply.github.com>
2024-08-12 17:01:41 +00:00
let value: &dyn PartialReflect = return_value.unwrap_ref();
assert_eq!(value.try_downcast_ref::<String>().unwrap(), "Hello, world!");
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
bevy_reflect: Function Overloading (Generic & Variadic Functions) (#15074) # Objective Currently function reflection requires users to manually monomorphize their generic functions. For example: ```rust fn add<T: Add<Output=T>>(a: T, b: T) -> T { a + b } // We have to specify the type of `T`: let reflect_add = add::<i32>.into_function(); ``` This PR doesn't aim to solve that problem—this is just a limitation in Rust. However, it also means that reflected functions can only ever work for a single monomorphization. If we wanted to support other types for `T`, we'd have to create a separate function for each one: ```rust let reflect_add_i32 = add::<i32>.into_function(); let reflect_add_u32 = add::<u32>.into_function(); let reflect_add_f32 = add::<f32>.into_function(); // ... ``` So in addition to requiring manual monomorphization, we also lose the benefit of having a single function handle multiple argument types. If a user wanted to create a small modding script that utilized function reflection, they'd have to either: - Store all sets of supported monomorphizations and require users to call the correct one - Write out some logic to find the correct function based on the given arguments While the first option would work, it wouldn't be very ergonomic. The second option is better, but it adds additional complexity to the user's logic—complexity that `bevy_reflect` could instead take on. ## Solution Introduce [function overloading](https://en.wikipedia.org/wiki/Function_overloading). A `DynamicFunction` can now be overloaded with other `DynamicFunction`s. We can rewrite the above code like so: ```rust let reflect_add = add::<i32> .into_function() .with_overload(add::<u32>) .with_overload(add::<f32>); ``` When invoked, the `DynamicFunction` will attempt to find a matching overload for the given set of arguments. And while I went into this PR only looking to improve generic function reflection, I accidentally added support for variadic functions as well (hence why I use the broader term "overload" over "generic"). ```rust // Supports 1 to 4 arguments let multiply_all = (|a: i32| a) .into_function() .with_overload(|a: i32, b: i32| a * b) .with_overload(|a: i32, b: i32, c: i32| a * b * c) .with_overload(|a: i32, b: i32, c: i32, d: i32| a * b * c * d); ``` This is simply an added bonus to this particular implementation. ~~Full variadic support (i.e. allowing for an indefinite number of arguments) will be added in a later PR.~~ I actually decided to limit the maximum number of arguments to 63 to supplement faster lookups, a reduced memory footprint, and faster cloning. ### Alternatives & Rationale I explored a few options for handling generic functions. This PR is the one I feel the most confident in, but I feel I should mention the others and why I ultimately didn't move forward with them. #### Adding `GenericDynamicFunction` **TL;DR:** Adding a distinct `GenericDynamicFunction` type unnecessarily splits and complicates the API. <details> <summary>Details</summary> My initial explorations involved a dedicated `GenericDynamicFunction` to contain and handle the mappings. This was initially started back when `DynamicFunction` was distinct from `DynamicClosure`. My goal was to not prevent us from being able to somehow make `DynamicFunction` implement `Copy`. But once we reverted back to a single `DynamicFunction`, that became a non-issue. But that aside, the real problem was that it created a split in the API. If I'm using a third-party library that uses function reflection, I have to know whether to request a `DynamicFunction` or a `GenericDynamicFunction`. I might not even know ahead of time which one I want. It might need to be determined at runtime. And if I'm creating a library, I might want a type to contain both `DynamicFunction` and `GenericDynamicFunction`. This might not be possible if, for example, I need to store the function in a `HashMap`. The other concern is with `IntoFunction`. Right now `DynamicFunction` trivially implements `IntoFunction` since it can just return itself. But what should `GenericDynamicFunction` do? It could return itself wrapped into a `DynamicFunction`, but then the API for `DynamicFunction` would have to account for this. So then what was the point of having a separate `GenericDynamicFunction` anyways? And even apart from `IntoFunction`, there's nothing stopping someone from manually creating a generic `DynamicFunction` through lying about its `FunctionInfo` and wrapping a `GenericDynamicFunction`. That being said, this is probably the "best" alternative if we added a `Function` trait and stored functions as `Box<dyn Function>`. However, I'm not convinced we gain much from this. Sure, we could keep the API for `DynamicFunction` the same, but consumers of `Function` will need to account for `GenericDynamicFunction` regardless (e.g. handling multiple `FunctionInfo`, a ranged argument count, etc.). And for all cases, except where using `DynamicFunction` directly, you end up treating them all like `GenericDynamicFunction`. Right now, if we did go with `GenericDynamicFunction`, the only major benefit we'd gain would be saving 24 bytes. If memory ever does become an issue here, we could swap over. But I think for the time being it's better for us to pursue a clearer mental model and end-user ergonomics through unification. </details> ##### Using the `FunctionRegistry` **TL;DR:** Having overloads only exist in the `FunctionRegistry` unnecessarily splits and complicates the API. <details> <summary>Details</summary> Another idea was to store the overloads in the `FunctionRegistry`. Users would then just call functions directly through the registry (i.e. `registry.call("my_func", my_args)`). I didn't go with this option because of how it specifically relies on the functions being registered. You'd not only always need access to the registry, but you'd need to ensure that the functions you want to call are even registered. It also means you can't just store a generic `DynamicFunction` on a type. Instead, you'll need to store the function's name and use that to look up the function in the registry—even if it's only ever used by that type. Doing so also removes all the benefits of `DynamicFunction`, such as the ability to pass it to functions accepting `IntoFunction`, modify it if needed, and so on. Like `GenericDynamicFunction` this introduces a split in the ecosystem: you either store `DynamicFunction`, store a string to look up the function, or force `DynamicFunction` to wrap your generic function anyways. Or worse yet: have `DynamicFunction` wrap the lookup function using `FunctionRegistryArc`. </details> #### Generic `ArgInfo` **TL;DR:** Allowing `ArgInfo` and `ReturnInfo` to store the generic information introduces a footgun when interpreting `FunctionInfo`. <details> <summary>Details</summary> Regardless of how we represent a generic function, one thing is clear: we need to be able to represent the information for such a function. This PR does so by introducing a `FunctionInfoType` enum to wrap one or more `FunctionInfo` values. Originally, I didn't do this. I had `ArgInfo` and `ReturnInfo` allow for generic types. This allowed us to have a single `FunctionInfo` to represent our function, but then I realized that it actually lies about our function. If we have two `ArgInfo` that both allow for either `i32` or `u32`, what does this tell us about our function? It turns out: nothing! We can't know whether our function takes `(i32, i32)`, `(u32, u32)`, `(i32, u32)`, or `(u32, i32)`. It therefore makes more sense to just represent a function with multiple `FunctionInfo` since that's really what it's made up of. </details> #### Flatten `FunctionInfo` **TL;DR:** Flattening removes additional per-overload information some users may desire and prevents us from adding more information in the future. <details> <summary>Details</summary> Why don't we just flatten multiple `FunctionInfo` into just one that can contain multiple signatures? This is something we could do, but I decided against it for a few reasons: - The only thing we'd be able to get rid of for each signature would be the `name`. While not enough to not do it, it doesn't really suggest we *have* to either. - Some consumers may want access to the names of the functions that make up the overloaded function. For example, to track a bug where an undesirable function is being added as an overload. Or to more easily locate the original function of an overload. - We may eventually allow for more information to be stored on `FunctionInfo`. For example, we may allow for documentation to be stored like we do for `TypeInfo`. Consumers of this documentation may want access to the documentation of each overload as they may provide documentation specific to that overload. </details> ## Testing This PR adds lots of tests and benchmarks, and also adds to the example. To run the tests: ``` cargo test --package bevy_reflect --all-features ``` To run the benchmarks: ``` cargo bench --bench reflect_function --all-features ``` To run the example: ``` cargo run --package bevy --example function_reflection --all-features ``` ### Benchmarks One of my goals with this PR was to leave the typical case of non-overloaded functions largely unaffected by the changes introduced in this PR. ~~And while the static size of `DynamicFunction` has increased by 17% (from 136 to 160 bytes), the performance has generally stayed the same~~ The static size of `DynamicFunction` has decreased from 136 to 112 bytes, while calling performance has generally stayed the same: | | `main` | 7d293ab | 252f3897d | |-------------------------------------|--------|---------|-----------| | `into/function` | 37 ns | 46 ns | 142 ns | | `with_overload/01_simple_overload` | - | 149 ns | 268 ns | | `with_overload/01_complex_overload` | - | 332 ns | 431 ns | | `with_overload/10_simple_overload` | - | 1266 ns | 2618 ns | | `with_overload/10_complex_overload` | - | 2544 ns | 4170 ns | | `call/function` | 57 ns | 58 ns | 61 ns | | `call/01_simple_overload` | - | 255 ns | 242 ns | | `call/01_complex_overload` | - | 595 ns | 431 ns | | `call/10_simple_overload` | - | 740 ns | 699 ns | | `call/10_complex_overload` | - | 1824 ns | 1618 ns | For the overloaded function tests, the leading number indicates how many overloads there are: `01` indicates 1 overload, `10` indicates 10 overloads. The `complex` cases have 10 unique generic types and 10 arguments, compared to the `simple` 1 generic type and 2 arguments. I aimed to prioritize the performance of calling the functions over creating them, hence creation speed tends to be a bit slower. There may be other optimizations we can look into but that's probably best saved for a future PR. The important bit is that the standard ~~`into/function`~~ and `call/function` benchmarks show minimal regressions. Since the latest changes, `into/function` does have some regressions, but again the priority was `call/function`. We can probably optimize `into/function` if needed in the future. --- ## Showcase Function reflection now supports [function overloading](https://en.wikipedia.org/wiki/Function_overloading)! This can be used to simulate generic functions: ```rust fn add<T: Add<Output=T>>(a: T, b: T) -> T { a + b } let reflect_add = add::<i32> .into_function() .with_overload(add::<u32>) .with_overload(add::<f32>); let args = ArgList::default().push_owned(25_i32).push_owned(75_i32); let result = func.call(args).unwrap().unwrap_owned(); assert_eq!(result.try_take::<i32>().unwrap(), 100); let args = ArgList::default().push_owned(25.0_f32).push_owned(75.0_f32); let result = func.call(args).unwrap().unwrap_owned(); assert_eq!(result.try_take::<f32>().unwrap(), 100.0); ``` You can also simulate variadic functions: ```rust #[derive(Reflect, PartialEq, Debug)] struct Player { name: Option<String>, health: u32, } // Creates a `Player` with one of the following: // - No name and 100 health // - A name and 100 health // - No name and custom health // - A name and custom health let create_player = (|| Player { name: None, health: 100, }) .into_function() .with_overload(|name: String| Player { name: Some(name), health: 100, }) .with_overload(|health: u32| Player { name: None, health }) .with_overload(|name: String, health: u32| Player { name: Some(name), health, }); let args = ArgList::default() .push_owned(String::from("Urist")) .push_owned(55_u32); let player = create_player .call(args) .unwrap() .unwrap_owned() .try_take::<Player>() .unwrap(); assert_eq!( player, Player { name: Some(String::from("Urist")), health: 55 } ); ```
2024-12-10 01:51:47 +00:00
// For more complex use cases, you can always create a custom `DynamicFunction` manually.
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
// This is useful for functions that can't be converted via the `IntoFunction` trait.
// For example, this function doesn't implement `IntoFunction` due to the fact that
// the lifetime of the return value is not tied to the lifetime of the first argument.
fn get_or_insert(value: i32, container: &mut Option<i32>) -> &i32 {
if container.is_none() {
*container = Some(value);
}
container.as_ref().unwrap()
}
bevy_reflect: Add `DynamicClosure` and `DynamicClosureMut` (#14141) # Objective As mentioned in [this](https://github.com/bevyengine/bevy/pull/13152#issuecomment-2198387297) comment, creating a function registry (see #14098) is a bit difficult due to the requirements of `DynamicFunction`. Internally, a `DynamicFunction` contains a `Box<dyn FnMut>` (the function that reifies reflected arguments and calls the actual function), which requires `&mut self` in order to be called. This means that users would require a mutable reference to the function registry for it to be useful— which isn't great. And they can't clone the `DynamicFunction` either because cloning an `FnMut` isn't really feasible (wrapping it in an `Arc` would allow it to be cloned but we wouldn't be able to call the clone since we need a mutable reference to the `FnMut`, which we can't get with multiple `Arc`s still alive, requiring us to also slap in a `Mutex`, which adds additional overhead). And we don't want to just replace the `dyn FnMut` with `dyn Fn` as that would prevent reflecting closures that mutate their environment. Instead, we need to introduce a new type to split the requirements of `DynamicFunction`. ## Solution Introduce new types for representing closures. Specifically, this PR introduces `DynamicClosure` and `DynamicClosureMut`. Similar to how `IntoFunction` exists for `DynamicFunction`, two new traits were introduced: `IntoClosure` and `IntoClosureMut`. Now `DynamicFunction` stores a `dyn Fn` with a `'static` lifetime. `DynamicClosure` also uses a `dyn Fn` but has a lifetime, `'env`, tied to its environment. `DynamicClosureMut` is most like the old `DynamicFunction`, keeping the `dyn FnMut` and also typing its lifetime, `'env`, to the environment Here are some comparison tables: | | `DynamicFunction` | `DynamicClosure` | `DynamicClosureMut` | | - | ----------------- | ---------------- | ------------------- | | Callable with `&self` | ✅ | ✅ | ❌ | | Callable with `&mut self` | ✅ | ✅ | ✅ | | Allows for non-`'static` lifetimes | ❌ | ✅ | ✅ | | | `IntoFunction` | `IntoClosure` | `IntoClosureMut` | | - | -------------- | ------------- | ---------------- | | Convert `fn` functions | ✅ | ✅ | ✅ | | Convert `fn` methods | ✅ | ✅ | ✅ | | Convert anonymous functions | ✅ | ✅ | ✅ | | Convert closures that capture immutable references | ❌ | ✅ | ✅ | | Convert closures that capture mutable references | ❌ | ❌ | ✅ | | Convert closures that capture owned values | ❌[^1] | ✅ | ✅ | [^1]: Due to limitations in Rust, `IntoFunction` can't be implemented for just functions (unless we forced users to manually coerce them to function pointers first). So closures that meet the trait requirements _can technically_ be converted into a `DynamicFunction` as well. To both future-proof and reduce confusion, though, we'll just pretend like this isn't a thing. ```rust let mut list: Vec<i32> = vec![1, 2, 3]; // `replace` is a closure that captures a mutable reference to `list` let mut replace = |index: usize, value: i32| -> i32 { let old_value = list[index]; list[index] = value; old_value }; // Convert the closure into a dynamic closure using `IntoClosureMut::into_closure_mut` let mut func: DynamicClosureMut = replace.into_closure_mut(); // Dynamically call the closure: let args = ArgList::default().push_owned(1_usize).push_owned(-2_i32); let value = func.call_once(args).unwrap().unwrap_owned(); // Check the result: assert_eq!(value.take::<i32>().unwrap(), 2); assert_eq!(list, vec![1, -2, 3]); ``` ### `ReflectFn`/`ReflectFnMut` To make extending the function reflection system easier (the blanket impls for `IntoFunction`, `IntoClosure`, and `IntoClosureMut` are all incredibly short), this PR generalizes callables with two new traits: `ReflectFn` and `ReflectFnMut`. These traits mimic `Fn` and `FnMut` but allow for being called via reflection. In fact, their blanket implementations are identical save for `ReflectFn` being implemented over `Fn` types and `ReflectFnMut` being implemented over `FnMut` types. And just as `Fn` is a subtrait of `FnMut`, `ReflectFn` is a subtrait of `ReflectFnMut`. So anywhere that expects a `ReflectFnMut` can also be given a `ReflectFn`. To reiterate, these traits aren't 100% necessary. They were added in purely for extensibility. If we decide to split things up differently or add new traits/types in the future, then those changes should be much simpler to implement. ### `TypedFunction` Because of the split into `ReflectFn` and `ReflectFnMut`, we needed a new way to access the function type information. This PR moves that concept over into `TypedFunction`. Much like `Typed`, this provides a way to access a function's `FunctionInfo`. By splitting this trait out, it helps to ensure the other traits are focused on a single responsibility. ### Internal Macros The original function PR (#13152) implemented `IntoFunction` using a macro which was passed into an `all_tuples!` macro invocation. Because we needed the same functionality for these new traits, this PR has copy+pasted that code for `ReflectFn`, `ReflectFnMut`, and `TypedFunction`— albeit with some differences between them. Originally, I was going to try and macro-ify the impls and where clauses such that we wouldn't have to straight up duplicate a lot of this logic. However, aside from being more complex in general, autocomplete just does not play nice with such heavily nested macros (tried in both RustRover and VSCode). And both of those problems told me that it just wasn't worth it: we need to ensure the crate is easily maintainable, even at the cost of duplicating code. So instead, I made sure to simplify the macro code by removing all fully-qualified syntax and cutting the where clauses down to the bare essentials, which helps to clean up a lot of the visual noise. I also tried my best to document the macro logic in certain areas (I may even add a bit more) to help with maintainability for future devs. ### Documentation Documentation for this module was a bit difficult for me. So many of these traits and types are very interconnected. And each trait/type has subtle differences that make documenting it in a single place, like at the module level, difficult to do cleanly. Describing the valid signatures is also challenging to do well. Hopefully what I have here is okay. I think I did an okay job, but let me know if there any thoughts on ways to improve it. We can also move such a task to a followup PR for more focused discussion. ## Testing You can test locally by running: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Added `DynamicClosure` struct - Added `DynamicClosureMut` struct - Added `IntoClosure` trait - Added `IntoClosureMut` trait - Added `ReflectFn` trait - Added `ReflectFnMut` trait - Added `TypedFunction` trait - `IntoFunction` now only works for standard Rust functions - `IntoFunction` no longer takes a lifetime parameter - `DynamicFunction::call` now only requires `&self` - Removed `DynamicFunction::call_once` - Changed the `IntoReturn::into_return` signature to include a where clause ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. ### `IntoClosure` `IntoFunction` now only works for standard Rust functions. Calling `IntoFunction::into_function` on a closure that captures references to its environment (either mutable or immutable), will no longer compile. Instead, you will need to use either `IntoClosure::into_closure` to create a `DynamicClosure` or `IntoClosureMut::into_closure_mut` to create a `DynamicClosureMut`, depending on your needs: ```rust let punct = String::from("!"); let print = |value: String| { println!("{value}{punct}"); }; // BEFORE let func: DynamicFunction = print.into_function(); // AFTER let func: DynamicClosure = print.into_closure(); ``` ### `IntoFunction` lifetime Additionally, `IntoFunction` no longer takes a lifetime parameter as it always expects a `'static` lifetime. Usages will need to remove any lifetime parameters: ```rust // BEFORE fn execute<'env, F: IntoFunction<'env, Marker>, Marker>(f: F) {/* ... */} // AFTER fn execute<F: IntoFunction<Marker>, Marker>(f: F) {/* ... */} ``` ### `IntoReturn` `IntoReturn::into_return` now has a where clause. Any manual implementors will need to add this where clause to their implementation.
2024-07-16 03:22:43 +00:00
let get_or_insert_function = dbg!(DynamicFunction::new(
|mut args: ArgList| -> FunctionResult {
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
// The `ArgList` contains the arguments in the order they were pushed.
// The `DynamicFunction` will validate that the list contains
// exactly the number of arguments we expect.
bevy_reflect: Improve `DynamicFunction` ergonomics (#14201) # Objective Many functions can be converted to `DynamicFunction` using `IntoFunction`. Unfortunately, we are limited by Rust itself and the implementations are far from exhaustive. For example, we can't convert functions with more than 16 arguments. Additionally, we can't handle returns with lifetimes not tied to the lifetime of the first argument. In such cases, users will have to create their `DynamicFunction` manually. Let's take the following function: ```rust fn get(index: usize, list: &Vec<String>) -> &String { &list[index] } ``` This function cannot be converted to a `DynamicFunction` via `IntoFunction` due to the lifetime of the return value being tied to the second argument. Therefore, we need to construct the `DynamicFunction` manually: ```rust DynamicFunction::new( |mut args, info| { let list = args .pop() .unwrap() .take_ref::<Vec<String>>(&info.args()[1])?; let index = args.pop().unwrap().take_owned::<usize>(&info.args()[0])?; Ok(Return::Ref(get(index, list))) }, FunctionInfo::new() .with_name("get") .with_args(vec![ ArgInfo::new::<usize>(0).with_name("index"), ArgInfo::new::<&Vec<String>>(1).with_name("list"), ]) .with_return_info(ReturnInfo::new::<&String>()), ); ``` While still a small and straightforward snippet, there's a decent amount going on here. There's a lot of room for improvements when it comes to ergonomics and readability. The goal of this PR is to address those issues. ## Solution Improve the ergonomics and readability of manually created `DynamicFunction`s. Some of the major changes: 1. Removed the need for `&ArgInfo` when reifying arguments (i.e. the `&info.args()[1]` calls) 2. Added additional `pop` methods on `ArgList` to handle both popping and casting 3. Added `take` methods on `ArgList` for taking the arguments out in order 4. Removed the need for `&FunctionInfo` in the internal closure (Change 1 made it no longer necessary) 5. Added methods to automatically handle generating `ArgInfo` and `ReturnInfo` With all these changes in place, we get something a lot nicer to both write and look at: ```rust DynamicFunction::new( |mut args| { let index = args.take::<usize>()?; let list = args.take::<&Vec<String>>()?; Ok(Return::Ref(get(index, list))) }, FunctionInfo::new() .with_name("get") .with_arg::<usize>("index") .with_arg::<&Vec<String>>("list") .with_return::<&String>(), ); ``` Alternatively, to rely on type inference for taking arguments, you could do: ```rust DynamicFunction::new( |mut args| { let index = args.take_owned()?; let list = args.take_ref()?; Ok(Return::Ref(get(index, list))) }, FunctionInfo::new() .with_name("get") .with_arg::<usize>("index") .with_arg::<&Vec<String>>("list") .with_return::<&String>(), ); ``` ## Testing You can test locally by running: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Removed `&ArgInfo` argument from `FromArg::from_arg` trait method - Removed `&ArgInfo` argument from `Arg::take_***` methods - Added `ArgValue` - `Arg` is now a struct containing an `ArgValue` and an argument `index` - `Arg::take_***` methods now require `T` is also `TypePath` - Added `Arg::new`, `Arg::index`, `Arg::value`, `Arg::take_value`, and `Arg::take` methods - Replaced `ArgId` in `ArgError` with just the argument `index` - Added `ArgError::EmptyArgList` - Renamed `ArgList::push` to `ArgList::push_arg` - Added `ArgList::pop_arg`, `ArgList::pop_owned`, `ArgList::pop_ref`, and `ArgList::pop_mut` - Added `ArgList::take_arg`, `ArgList::take_owned`, `ArgList::take_ref`, `ArgList::take_mut`, and `ArgList::take` - `ArgList::pop` is now generic - Renamed `FunctionError::InvalidArgCount` to `FunctionError::ArgCountMismatch` - The closure given to `DynamicFunction::new` no longer has a `&FunctionInfo` argument - Added `FunctionInfo::with_arg` - Added `FunctionInfo::with_return` ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. * The `FromArg::from_arg` trait method and the `Arg::take_***` methods no longer take a `&ArgInfo` argument. * What used to be `Arg` is now `ArgValue`. `Arg` is now a struct which contains an `ArgValue`. * `Arg::take_***` methods now require `T` is also `TypePath` * Instances of `id: ArgId` in `ArgError` have been replaced with `index: usize` * `ArgList::push` is now `ArgList::push_arg`. It also takes the new `ArgValue` type. * `ArgList::pop` has become `ArgList::pop_arg` and now returns `ArgValue`. `Arg::pop` now takes a generic type and downcasts to that type. It's recommended to use `ArgList::take` and friends instead since they allow removing the arguments from the list in the order they were pushed (rather than reverse order). * `FunctionError::InvalidArgCount` is now `FunctionError::ArgCountMismatch` * The closure given to `DynamicFunction::new` no longer has a `&FunctionInfo` argument. This argument can be removed.
2024-07-16 13:01:52 +00:00
// We can retrieve them out in order (note that this modifies the `ArgList`):
let value = args.take::<i32>()?;
let container = args.take::<&mut Option<i32>>()?;
// We could have also done the following to make use of type inference:
// let value = args.take_owned()?;
// let container = args.take_mut()?;
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
Ok(Return::Ref(get_or_insert(value, container)))
},
bevy_reflect: Anonymous function parsing (#14641) # Objective ### TL;DR #14098 added the `FunctionRegistry` but had some last minute complications due to anonymous functions. It ended up going with a "required name" approach to ensure anonymous functions would always have a name. However, this approach isn't ideal for named functions since, by definition, they will always have a name. Therefore, this PR aims to modify function reflection such that we can make function registration easier for named functions, while still allowing anonymous functions to be registered as well. ### Context Function registration (#14098) ran into a little problem: anonymous functions. Anonymous functions, including function pointers, have very non-unique type names. For example, the anonymous function `|a: i32, b: i32| a + b` has the type name of `fn(i32, i32) -> i32`. This obviously means we'd conflict with another function like `|a: i32, b: i32| a - b`. The solution that #14098 landed on was to always require a name during function registration. The downside with this is that named functions (e.g. `fn add(a: i32, b: i32) -> i32 { a + b }`) had to redundantly provide a name. Additionally, manually constructed `DynamicFunction`s also ran into this ergonomics issue. I don't entirely know how the function registry will be used, but I have a strong suspicion that most of its registrations will either be named functions or manually constructed `DynamicFunction`s, with anonymous functions only being used here and there for quick prototyping or adding small functionality. Why then should the API prioritize the anonymous function use case by always requiring a name during registration? #### Telling Functions Apart Rust doesn't provide a lot of out-of-the-box tools for reflecting functions. One of the biggest hurdles in attempting to solve the problem outlined above would be to somehow tell the different kinds of functions apart. Let's briefly recap on the categories of functions in Rust: | Category | Example | | ------------------ | ----------------------------------------- | | Named function | `fn add(a: i32, b: i32) -> i32 { a + b }` | | Closure | `\|a: i32\| a + captured_variable` | | Anonymous function | `\|a: i32, b: i32\| a + b` | | Function pointer | `fn(i32, i32) -> i32` | My first thought was to try and differentiate these categories based on their size. However, we can see that this doesn't quite work: | Category | `size_of` | | ------------------ | --------- | | Named function | 0 | | Closure | 0+ | | Anonymous function | 0 | | Function pointer | 8 | Not only does this not tell anonymous functions from named ones, but it struggles with pretty much all of them. My second then was to differentiate based on type name: | Category | `type_name` | | ------------------ | ----------------------- | | Named function | `foo::bar::baz` | | Closure | `foo::bar::{{closure}}` | | Anonymous function | `fn() -> String` | | Function pointer | `fn() -> String` | This is much better. While it can't distinguish between function pointers and anonymous functions, this doesn't matter too much since we only care about whether we can _name_ the function. So why didn't we implement this in #14098? #### Relying on `type_name` While this solution was known about while working on #14098, it was left out from that PR due to it being potentially controversial. The [docs](https://doc.rust-lang.org/stable/std/any/fn.type_name.html) for `std::any::type_name` state: > The returned string must not be considered to be a unique identifier of a type as multiple types may map to the same type name. Similarly, there is no guarantee that all parts of a type will appear in the returned string: for example, lifetime specifiers are currently not included. In addition, the output may change between versions of the compiler. So that's it then? We can't use `type_name`? Well, this statement isn't so much a rule as it is a guideline. And Bevy is no stranger to bending the rules to make things work or to improve ergonomics. Remember that before `TypePath`, Bevy's scene system was entirely dependent on `type_name`. Not to mention that `type_name` is being used as a key into both the `TypeRegistry` and the `FunctionRegistry`. Bevy's practices aside, can we reliably use `type_name` for this? My answer would be "yes". Anonymous functions are anonymous. They have no name. There's nothing Rust could do to give them a name apart from generating a random string of characters. But remember that this is a diagnostic tool, it doesn't make sense to obfuscate the type by randomizing the output. So changing it to be anything other than what it is now is very unlikely. The only changes that I could potentially see happening are: 1. Closures replace `{{closure}}` with the name of their variable 2. Lifetimes are included in the output I don't think the first is likely to happen, but if it does then it actually works out in our favor: closures are now named! The second point is probably the likeliest. However, adding lifetimes doesn't mean we can't still rely on `type_name` to determine whether or not a function is named. So we should be okay in this case as well. ## Solution Parse the `type_name` of the function in the `TypedFunction` impl to determine if the function is named or anonymous. This once again makes `FunctionInfo::name` optional. For manual constructions of `DynamicFunction`, `FunctionInfo::named` or ``FunctionInfo::anonymous` can be used. The `FunctionRegistry` API has also been reworked to account for this change. `FunctionRegistry::register` no longer takes a name and instead takes it from the supplied function, returning a `FunctionRegistrationError::MissingName` error if the name is `None`. This also doubles as a replacement for the old `FunctionRegistry::register_dynamic` method, which has been removed. To handle anonymous functions, a `FunctionRegistry::register_with_name` method has been added. This works in the same way `FunctionRegistry::register` used to work before this PR. The overwriting methods have been updated in a similar manner, with modifications to `FunctionRegistry::overwrite_registration`, the removal of `FunctionRegistry::overwrite_registration_dynamic`, and the addition of `FunctionRegistry::overwrite_registration_with_name`. This PR also updates the methods on `App` in a similar way: `App::register_function` no longer requires a name argument and `App::register_function_with_name` has been added to handle anonymous functions (and eventually closures). ## Testing You can run the tests locally by running: ``` cargo test --package bevy_reflect --features functions ``` --- ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. > [!note] > This list is not exhaustive. It only contains some of the most important changes. `FunctionRegistry::register` no longer requires a name string for named functions. Anonymous functions, however, need to be registered using `FunctionRegistry::register_with_name`. ```rust // BEFORE registry .register(std::any::type_name_of_val(&foo), foo)? .register("bar", || println!("Hello world!")); // AFTER registry .register(foo)? .register_with_name("bar", || println!("Hello world!")); ``` `FunctionInfo::name` is now optional. Anonymous functions and closures will now have their name set to `None` by default. Additionally, `FunctionInfo::new` has been renamed to `FunctionInfo::named`.
2024-08-07 03:11:08 +00:00
// Functions can be either anonymous or named.
// It's good practice, though, to try and name your functions whenever possible.
// This makes it easier to debug and is also required for function registration.
// We can either give it a custom name or use the function's type name as
bevy_reflect: Function registry (#14098) # Objective #13152 added support for reflecting functions. Now, we need a way to register those functions such that they may be accessed anywhere within the ECS. ## Solution Added a `FunctionRegistry` type similar to `TypeRegistry`. This allows a function to be registered and retrieved by name. ```rust fn foo() -> i32 { 123 } let mut registry = FunctionRegistry::default(); registry.register("my_function", foo); let function = registry.get_mut("my_function").unwrap(); let value = function.call(ArgList::new()).unwrap().unwrap_owned(); assert_eq!(value.downcast_ref::<i32>(), Some(&123)); ``` Additionally, I added an `AppFunctionRegistry` resource which wraps a `FunctionRegistryArc`. Functions can be registered into this resource using `App::register_function` or by getting a mutable reference to the resource itself. ### Limitations #### `Send + Sync` In order to get this registry to work across threads, it needs to be `Send + Sync`. This means that `DynamicFunction` needs to be `Send + Sync`, which means that its internal function also needs to be `Send + Sync`. In most cases, this won't be an issue because standard Rust functions (the type most likely to be registered) are always `Send + Sync`. Additionally, closures tend to be `Send + Sync` as well, granted they don't capture any `!Send` or `!Sync` variables. This PR adds this `Send + Sync` requirement, but as mentioned above, it hopefully shouldn't be too big of an issue. #### Closures Unfortunately, closures can't be registered yet. This will likely be explored and added in a followup PR. ### Future Work Besides addressing the limitations listed above, another thing we could look into is improving the lookup of registered functions. One aspect is in the performance of hashing strings. The other is in the developer experience of having to call `std::any::type_name_of_val` to get the name of their function (assuming they didn't give it a custom name). ## Testing You can run the tests locally with: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Added `FunctionRegistry` - Added `AppFunctionRegistry` (a `Resource` available from `bevy_ecs`) - Added `FunctionRegistryArc` - Added `FunctionRegistrationError` - Added `reflect_functions` feature to `bevy_ecs` and `bevy_app` - `FunctionInfo` is no longer `Default` - `DynamicFunction` now requires its wrapped function be `Send + Sync` ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. `DynamicFunction` (both those created manually and those created with `IntoFunction`), now require `Send + Sync`. All standard Rust functions should meet that requirement. Closures, on the other hand, may not if they capture any `!Send` or `!Sync` variables from its environment.
2024-08-06 01:09:48 +00:00
// derived from `std::any::type_name_of_val`.
bevy_reflect: Function Overloading (Generic & Variadic Functions) (#15074) # Objective Currently function reflection requires users to manually monomorphize their generic functions. For example: ```rust fn add<T: Add<Output=T>>(a: T, b: T) -> T { a + b } // We have to specify the type of `T`: let reflect_add = add::<i32>.into_function(); ``` This PR doesn't aim to solve that problem—this is just a limitation in Rust. However, it also means that reflected functions can only ever work for a single monomorphization. If we wanted to support other types for `T`, we'd have to create a separate function for each one: ```rust let reflect_add_i32 = add::<i32>.into_function(); let reflect_add_u32 = add::<u32>.into_function(); let reflect_add_f32 = add::<f32>.into_function(); // ... ``` So in addition to requiring manual monomorphization, we also lose the benefit of having a single function handle multiple argument types. If a user wanted to create a small modding script that utilized function reflection, they'd have to either: - Store all sets of supported monomorphizations and require users to call the correct one - Write out some logic to find the correct function based on the given arguments While the first option would work, it wouldn't be very ergonomic. The second option is better, but it adds additional complexity to the user's logic—complexity that `bevy_reflect` could instead take on. ## Solution Introduce [function overloading](https://en.wikipedia.org/wiki/Function_overloading). A `DynamicFunction` can now be overloaded with other `DynamicFunction`s. We can rewrite the above code like so: ```rust let reflect_add = add::<i32> .into_function() .with_overload(add::<u32>) .with_overload(add::<f32>); ``` When invoked, the `DynamicFunction` will attempt to find a matching overload for the given set of arguments. And while I went into this PR only looking to improve generic function reflection, I accidentally added support for variadic functions as well (hence why I use the broader term "overload" over "generic"). ```rust // Supports 1 to 4 arguments let multiply_all = (|a: i32| a) .into_function() .with_overload(|a: i32, b: i32| a * b) .with_overload(|a: i32, b: i32, c: i32| a * b * c) .with_overload(|a: i32, b: i32, c: i32, d: i32| a * b * c * d); ``` This is simply an added bonus to this particular implementation. ~~Full variadic support (i.e. allowing for an indefinite number of arguments) will be added in a later PR.~~ I actually decided to limit the maximum number of arguments to 63 to supplement faster lookups, a reduced memory footprint, and faster cloning. ### Alternatives & Rationale I explored a few options for handling generic functions. This PR is the one I feel the most confident in, but I feel I should mention the others and why I ultimately didn't move forward with them. #### Adding `GenericDynamicFunction` **TL;DR:** Adding a distinct `GenericDynamicFunction` type unnecessarily splits and complicates the API. <details> <summary>Details</summary> My initial explorations involved a dedicated `GenericDynamicFunction` to contain and handle the mappings. This was initially started back when `DynamicFunction` was distinct from `DynamicClosure`. My goal was to not prevent us from being able to somehow make `DynamicFunction` implement `Copy`. But once we reverted back to a single `DynamicFunction`, that became a non-issue. But that aside, the real problem was that it created a split in the API. If I'm using a third-party library that uses function reflection, I have to know whether to request a `DynamicFunction` or a `GenericDynamicFunction`. I might not even know ahead of time which one I want. It might need to be determined at runtime. And if I'm creating a library, I might want a type to contain both `DynamicFunction` and `GenericDynamicFunction`. This might not be possible if, for example, I need to store the function in a `HashMap`. The other concern is with `IntoFunction`. Right now `DynamicFunction` trivially implements `IntoFunction` since it can just return itself. But what should `GenericDynamicFunction` do? It could return itself wrapped into a `DynamicFunction`, but then the API for `DynamicFunction` would have to account for this. So then what was the point of having a separate `GenericDynamicFunction` anyways? And even apart from `IntoFunction`, there's nothing stopping someone from manually creating a generic `DynamicFunction` through lying about its `FunctionInfo` and wrapping a `GenericDynamicFunction`. That being said, this is probably the "best" alternative if we added a `Function` trait and stored functions as `Box<dyn Function>`. However, I'm not convinced we gain much from this. Sure, we could keep the API for `DynamicFunction` the same, but consumers of `Function` will need to account for `GenericDynamicFunction` regardless (e.g. handling multiple `FunctionInfo`, a ranged argument count, etc.). And for all cases, except where using `DynamicFunction` directly, you end up treating them all like `GenericDynamicFunction`. Right now, if we did go with `GenericDynamicFunction`, the only major benefit we'd gain would be saving 24 bytes. If memory ever does become an issue here, we could swap over. But I think for the time being it's better for us to pursue a clearer mental model and end-user ergonomics through unification. </details> ##### Using the `FunctionRegistry` **TL;DR:** Having overloads only exist in the `FunctionRegistry` unnecessarily splits and complicates the API. <details> <summary>Details</summary> Another idea was to store the overloads in the `FunctionRegistry`. Users would then just call functions directly through the registry (i.e. `registry.call("my_func", my_args)`). I didn't go with this option because of how it specifically relies on the functions being registered. You'd not only always need access to the registry, but you'd need to ensure that the functions you want to call are even registered. It also means you can't just store a generic `DynamicFunction` on a type. Instead, you'll need to store the function's name and use that to look up the function in the registry—even if it's only ever used by that type. Doing so also removes all the benefits of `DynamicFunction`, such as the ability to pass it to functions accepting `IntoFunction`, modify it if needed, and so on. Like `GenericDynamicFunction` this introduces a split in the ecosystem: you either store `DynamicFunction`, store a string to look up the function, or force `DynamicFunction` to wrap your generic function anyways. Or worse yet: have `DynamicFunction` wrap the lookup function using `FunctionRegistryArc`. </details> #### Generic `ArgInfo` **TL;DR:** Allowing `ArgInfo` and `ReturnInfo` to store the generic information introduces a footgun when interpreting `FunctionInfo`. <details> <summary>Details</summary> Regardless of how we represent a generic function, one thing is clear: we need to be able to represent the information for such a function. This PR does so by introducing a `FunctionInfoType` enum to wrap one or more `FunctionInfo` values. Originally, I didn't do this. I had `ArgInfo` and `ReturnInfo` allow for generic types. This allowed us to have a single `FunctionInfo` to represent our function, but then I realized that it actually lies about our function. If we have two `ArgInfo` that both allow for either `i32` or `u32`, what does this tell us about our function? It turns out: nothing! We can't know whether our function takes `(i32, i32)`, `(u32, u32)`, `(i32, u32)`, or `(u32, i32)`. It therefore makes more sense to just represent a function with multiple `FunctionInfo` since that's really what it's made up of. </details> #### Flatten `FunctionInfo` **TL;DR:** Flattening removes additional per-overload information some users may desire and prevents us from adding more information in the future. <details> <summary>Details</summary> Why don't we just flatten multiple `FunctionInfo` into just one that can contain multiple signatures? This is something we could do, but I decided against it for a few reasons: - The only thing we'd be able to get rid of for each signature would be the `name`. While not enough to not do it, it doesn't really suggest we *have* to either. - Some consumers may want access to the names of the functions that make up the overloaded function. For example, to track a bug where an undesirable function is being added as an overload. Or to more easily locate the original function of an overload. - We may eventually allow for more information to be stored on `FunctionInfo`. For example, we may allow for documentation to be stored like we do for `TypeInfo`. Consumers of this documentation may want access to the documentation of each overload as they may provide documentation specific to that overload. </details> ## Testing This PR adds lots of tests and benchmarks, and also adds to the example. To run the tests: ``` cargo test --package bevy_reflect --all-features ``` To run the benchmarks: ``` cargo bench --bench reflect_function --all-features ``` To run the example: ``` cargo run --package bevy --example function_reflection --all-features ``` ### Benchmarks One of my goals with this PR was to leave the typical case of non-overloaded functions largely unaffected by the changes introduced in this PR. ~~And while the static size of `DynamicFunction` has increased by 17% (from 136 to 160 bytes), the performance has generally stayed the same~~ The static size of `DynamicFunction` has decreased from 136 to 112 bytes, while calling performance has generally stayed the same: | | `main` | 7d293ab | 252f3897d | |-------------------------------------|--------|---------|-----------| | `into/function` | 37 ns | 46 ns | 142 ns | | `with_overload/01_simple_overload` | - | 149 ns | 268 ns | | `with_overload/01_complex_overload` | - | 332 ns | 431 ns | | `with_overload/10_simple_overload` | - | 1266 ns | 2618 ns | | `with_overload/10_complex_overload` | - | 2544 ns | 4170 ns | | `call/function` | 57 ns | 58 ns | 61 ns | | `call/01_simple_overload` | - | 255 ns | 242 ns | | `call/01_complex_overload` | - | 595 ns | 431 ns | | `call/10_simple_overload` | - | 740 ns | 699 ns | | `call/10_complex_overload` | - | 1824 ns | 1618 ns | For the overloaded function tests, the leading number indicates how many overloads there are: `01` indicates 1 overload, `10` indicates 10 overloads. The `complex` cases have 10 unique generic types and 10 arguments, compared to the `simple` 1 generic type and 2 arguments. I aimed to prioritize the performance of calling the functions over creating them, hence creation speed tends to be a bit slower. There may be other optimizations we can look into but that's probably best saved for a future PR. The important bit is that the standard ~~`into/function`~~ and `call/function` benchmarks show minimal regressions. Since the latest changes, `into/function` does have some regressions, but again the priority was `call/function`. We can probably optimize `into/function` if needed in the future. --- ## Showcase Function reflection now supports [function overloading](https://en.wikipedia.org/wiki/Function_overloading)! This can be used to simulate generic functions: ```rust fn add<T: Add<Output=T>>(a: T, b: T) -> T { a + b } let reflect_add = add::<i32> .into_function() .with_overload(add::<u32>) .with_overload(add::<f32>); let args = ArgList::default().push_owned(25_i32).push_owned(75_i32); let result = func.call(args).unwrap().unwrap_owned(); assert_eq!(result.try_take::<i32>().unwrap(), 100); let args = ArgList::default().push_owned(25.0_f32).push_owned(75.0_f32); let result = func.call(args).unwrap().unwrap_owned(); assert_eq!(result.try_take::<f32>().unwrap(), 100.0); ``` You can also simulate variadic functions: ```rust #[derive(Reflect, PartialEq, Debug)] struct Player { name: Option<String>, health: u32, } // Creates a `Player` with one of the following: // - No name and 100 health // - A name and 100 health // - No name and custom health // - A name and custom health let create_player = (|| Player { name: None, health: 100, }) .into_function() .with_overload(|name: String| Player { name: Some(name), health: 100, }) .with_overload(|health: u32| Player { name: None, health }) .with_overload(|name: String, health: u32| Player { name: Some(name), health, }); let args = ArgList::default() .push_owned(String::from("Urist")) .push_owned(55_u32); let player = create_player .call(args) .unwrap() .unwrap_owned() .try_take::<Player>() .unwrap(); assert_eq!( player, Player { name: Some(String::from("Urist")), health: 55 } ); ```
2024-12-10 01:51:47 +00:00
SignatureInfo::named(std::any::type_name_of_val(&get_or_insert))
bevy_reflect: Function registry (#14098) # Objective #13152 added support for reflecting functions. Now, we need a way to register those functions such that they may be accessed anywhere within the ECS. ## Solution Added a `FunctionRegistry` type similar to `TypeRegistry`. This allows a function to be registered and retrieved by name. ```rust fn foo() -> i32 { 123 } let mut registry = FunctionRegistry::default(); registry.register("my_function", foo); let function = registry.get_mut("my_function").unwrap(); let value = function.call(ArgList::new()).unwrap().unwrap_owned(); assert_eq!(value.downcast_ref::<i32>(), Some(&123)); ``` Additionally, I added an `AppFunctionRegistry` resource which wraps a `FunctionRegistryArc`. Functions can be registered into this resource using `App::register_function` or by getting a mutable reference to the resource itself. ### Limitations #### `Send + Sync` In order to get this registry to work across threads, it needs to be `Send + Sync`. This means that `DynamicFunction` needs to be `Send + Sync`, which means that its internal function also needs to be `Send + Sync`. In most cases, this won't be an issue because standard Rust functions (the type most likely to be registered) are always `Send + Sync`. Additionally, closures tend to be `Send + Sync` as well, granted they don't capture any `!Send` or `!Sync` variables. This PR adds this `Send + Sync` requirement, but as mentioned above, it hopefully shouldn't be too big of an issue. #### Closures Unfortunately, closures can't be registered yet. This will likely be explored and added in a followup PR. ### Future Work Besides addressing the limitations listed above, another thing we could look into is improving the lookup of registered functions. One aspect is in the performance of hashing strings. The other is in the developer experience of having to call `std::any::type_name_of_val` to get the name of their function (assuming they didn't give it a custom name). ## Testing You can run the tests locally with: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Added `FunctionRegistry` - Added `AppFunctionRegistry` (a `Resource` available from `bevy_ecs`) - Added `FunctionRegistryArc` - Added `FunctionRegistrationError` - Added `reflect_functions` feature to `bevy_ecs` and `bevy_app` - `FunctionInfo` is no longer `Default` - `DynamicFunction` now requires its wrapped function be `Send + Sync` ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. `DynamicFunction` (both those created manually and those created with `IntoFunction`), now require `Send + Sync`. All standard Rust functions should meet that requirement. Closures, on the other hand, may not if they capture any `!Send` or `!Sync` variables from its environment.
2024-08-06 01:09:48 +00:00
// We can always change the name if needed.
bevy_reflect: Anonymous function parsing (#14641) # Objective ### TL;DR #14098 added the `FunctionRegistry` but had some last minute complications due to anonymous functions. It ended up going with a "required name" approach to ensure anonymous functions would always have a name. However, this approach isn't ideal for named functions since, by definition, they will always have a name. Therefore, this PR aims to modify function reflection such that we can make function registration easier for named functions, while still allowing anonymous functions to be registered as well. ### Context Function registration (#14098) ran into a little problem: anonymous functions. Anonymous functions, including function pointers, have very non-unique type names. For example, the anonymous function `|a: i32, b: i32| a + b` has the type name of `fn(i32, i32) -> i32`. This obviously means we'd conflict with another function like `|a: i32, b: i32| a - b`. The solution that #14098 landed on was to always require a name during function registration. The downside with this is that named functions (e.g. `fn add(a: i32, b: i32) -> i32 { a + b }`) had to redundantly provide a name. Additionally, manually constructed `DynamicFunction`s also ran into this ergonomics issue. I don't entirely know how the function registry will be used, but I have a strong suspicion that most of its registrations will either be named functions or manually constructed `DynamicFunction`s, with anonymous functions only being used here and there for quick prototyping or adding small functionality. Why then should the API prioritize the anonymous function use case by always requiring a name during registration? #### Telling Functions Apart Rust doesn't provide a lot of out-of-the-box tools for reflecting functions. One of the biggest hurdles in attempting to solve the problem outlined above would be to somehow tell the different kinds of functions apart. Let's briefly recap on the categories of functions in Rust: | Category | Example | | ------------------ | ----------------------------------------- | | Named function | `fn add(a: i32, b: i32) -> i32 { a + b }` | | Closure | `\|a: i32\| a + captured_variable` | | Anonymous function | `\|a: i32, b: i32\| a + b` | | Function pointer | `fn(i32, i32) -> i32` | My first thought was to try and differentiate these categories based on their size. However, we can see that this doesn't quite work: | Category | `size_of` | | ------------------ | --------- | | Named function | 0 | | Closure | 0+ | | Anonymous function | 0 | | Function pointer | 8 | Not only does this not tell anonymous functions from named ones, but it struggles with pretty much all of them. My second then was to differentiate based on type name: | Category | `type_name` | | ------------------ | ----------------------- | | Named function | `foo::bar::baz` | | Closure | `foo::bar::{{closure}}` | | Anonymous function | `fn() -> String` | | Function pointer | `fn() -> String` | This is much better. While it can't distinguish between function pointers and anonymous functions, this doesn't matter too much since we only care about whether we can _name_ the function. So why didn't we implement this in #14098? #### Relying on `type_name` While this solution was known about while working on #14098, it was left out from that PR due to it being potentially controversial. The [docs](https://doc.rust-lang.org/stable/std/any/fn.type_name.html) for `std::any::type_name` state: > The returned string must not be considered to be a unique identifier of a type as multiple types may map to the same type name. Similarly, there is no guarantee that all parts of a type will appear in the returned string: for example, lifetime specifiers are currently not included. In addition, the output may change between versions of the compiler. So that's it then? We can't use `type_name`? Well, this statement isn't so much a rule as it is a guideline. And Bevy is no stranger to bending the rules to make things work or to improve ergonomics. Remember that before `TypePath`, Bevy's scene system was entirely dependent on `type_name`. Not to mention that `type_name` is being used as a key into both the `TypeRegistry` and the `FunctionRegistry`. Bevy's practices aside, can we reliably use `type_name` for this? My answer would be "yes". Anonymous functions are anonymous. They have no name. There's nothing Rust could do to give them a name apart from generating a random string of characters. But remember that this is a diagnostic tool, it doesn't make sense to obfuscate the type by randomizing the output. So changing it to be anything other than what it is now is very unlikely. The only changes that I could potentially see happening are: 1. Closures replace `{{closure}}` with the name of their variable 2. Lifetimes are included in the output I don't think the first is likely to happen, but if it does then it actually works out in our favor: closures are now named! The second point is probably the likeliest. However, adding lifetimes doesn't mean we can't still rely on `type_name` to determine whether or not a function is named. So we should be okay in this case as well. ## Solution Parse the `type_name` of the function in the `TypedFunction` impl to determine if the function is named or anonymous. This once again makes `FunctionInfo::name` optional. For manual constructions of `DynamicFunction`, `FunctionInfo::named` or ``FunctionInfo::anonymous` can be used. The `FunctionRegistry` API has also been reworked to account for this change. `FunctionRegistry::register` no longer takes a name and instead takes it from the supplied function, returning a `FunctionRegistrationError::MissingName` error if the name is `None`. This also doubles as a replacement for the old `FunctionRegistry::register_dynamic` method, which has been removed. To handle anonymous functions, a `FunctionRegistry::register_with_name` method has been added. This works in the same way `FunctionRegistry::register` used to work before this PR. The overwriting methods have been updated in a similar manner, with modifications to `FunctionRegistry::overwrite_registration`, the removal of `FunctionRegistry::overwrite_registration_dynamic`, and the addition of `FunctionRegistry::overwrite_registration_with_name`. This PR also updates the methods on `App` in a similar way: `App::register_function` no longer requires a name argument and `App::register_function_with_name` has been added to handle anonymous functions (and eventually closures). ## Testing You can run the tests locally by running: ``` cargo test --package bevy_reflect --features functions ``` --- ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. > [!note] > This list is not exhaustive. It only contains some of the most important changes. `FunctionRegistry::register` no longer requires a name string for named functions. Anonymous functions, however, need to be registered using `FunctionRegistry::register_with_name`. ```rust // BEFORE registry .register(std::any::type_name_of_val(&foo), foo)? .register("bar", || println!("Hello world!")); // AFTER registry .register(foo)? .register_with_name("bar", || println!("Hello world!")); ``` `FunctionInfo::name` is now optional. Anonymous functions and closures will now have their name set to `None` by default. Additionally, `FunctionInfo::new` has been renamed to `FunctionInfo::named`.
2024-08-07 03:11:08 +00:00
// It's a good idea to also ensure that the name is unique,
// such as by using its type name or by prefixing it with your crate name.
.with_name("my_crate::get_or_insert")
bevy_reflect: Improve `DynamicFunction` ergonomics (#14201) # Objective Many functions can be converted to `DynamicFunction` using `IntoFunction`. Unfortunately, we are limited by Rust itself and the implementations are far from exhaustive. For example, we can't convert functions with more than 16 arguments. Additionally, we can't handle returns with lifetimes not tied to the lifetime of the first argument. In such cases, users will have to create their `DynamicFunction` manually. Let's take the following function: ```rust fn get(index: usize, list: &Vec<String>) -> &String { &list[index] } ``` This function cannot be converted to a `DynamicFunction` via `IntoFunction` due to the lifetime of the return value being tied to the second argument. Therefore, we need to construct the `DynamicFunction` manually: ```rust DynamicFunction::new( |mut args, info| { let list = args .pop() .unwrap() .take_ref::<Vec<String>>(&info.args()[1])?; let index = args.pop().unwrap().take_owned::<usize>(&info.args()[0])?; Ok(Return::Ref(get(index, list))) }, FunctionInfo::new() .with_name("get") .with_args(vec![ ArgInfo::new::<usize>(0).with_name("index"), ArgInfo::new::<&Vec<String>>(1).with_name("list"), ]) .with_return_info(ReturnInfo::new::<&String>()), ); ``` While still a small and straightforward snippet, there's a decent amount going on here. There's a lot of room for improvements when it comes to ergonomics and readability. The goal of this PR is to address those issues. ## Solution Improve the ergonomics and readability of manually created `DynamicFunction`s. Some of the major changes: 1. Removed the need for `&ArgInfo` when reifying arguments (i.e. the `&info.args()[1]` calls) 2. Added additional `pop` methods on `ArgList` to handle both popping and casting 3. Added `take` methods on `ArgList` for taking the arguments out in order 4. Removed the need for `&FunctionInfo` in the internal closure (Change 1 made it no longer necessary) 5. Added methods to automatically handle generating `ArgInfo` and `ReturnInfo` With all these changes in place, we get something a lot nicer to both write and look at: ```rust DynamicFunction::new( |mut args| { let index = args.take::<usize>()?; let list = args.take::<&Vec<String>>()?; Ok(Return::Ref(get(index, list))) }, FunctionInfo::new() .with_name("get") .with_arg::<usize>("index") .with_arg::<&Vec<String>>("list") .with_return::<&String>(), ); ``` Alternatively, to rely on type inference for taking arguments, you could do: ```rust DynamicFunction::new( |mut args| { let index = args.take_owned()?; let list = args.take_ref()?; Ok(Return::Ref(get(index, list))) }, FunctionInfo::new() .with_name("get") .with_arg::<usize>("index") .with_arg::<&Vec<String>>("list") .with_return::<&String>(), ); ``` ## Testing You can test locally by running: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Removed `&ArgInfo` argument from `FromArg::from_arg` trait method - Removed `&ArgInfo` argument from `Arg::take_***` methods - Added `ArgValue` - `Arg` is now a struct containing an `ArgValue` and an argument `index` - `Arg::take_***` methods now require `T` is also `TypePath` - Added `Arg::new`, `Arg::index`, `Arg::value`, `Arg::take_value`, and `Arg::take` methods - Replaced `ArgId` in `ArgError` with just the argument `index` - Added `ArgError::EmptyArgList` - Renamed `ArgList::push` to `ArgList::push_arg` - Added `ArgList::pop_arg`, `ArgList::pop_owned`, `ArgList::pop_ref`, and `ArgList::pop_mut` - Added `ArgList::take_arg`, `ArgList::take_owned`, `ArgList::take_ref`, `ArgList::take_mut`, and `ArgList::take` - `ArgList::pop` is now generic - Renamed `FunctionError::InvalidArgCount` to `FunctionError::ArgCountMismatch` - The closure given to `DynamicFunction::new` no longer has a `&FunctionInfo` argument - Added `FunctionInfo::with_arg` - Added `FunctionInfo::with_return` ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. * The `FromArg::from_arg` trait method and the `Arg::take_***` methods no longer take a `&ArgInfo` argument. * What used to be `Arg` is now `ArgValue`. `Arg` is now a struct which contains an `ArgValue`. * `Arg::take_***` methods now require `T` is also `TypePath` * Instances of `id: ArgId` in `ArgError` have been replaced with `index: usize` * `ArgList::push` is now `ArgList::push_arg`. It also takes the new `ArgValue` type. * `ArgList::pop` has become `ArgList::pop_arg` and now returns `ArgValue`. `Arg::pop` now takes a generic type and downcasts to that type. It's recommended to use `ArgList::take` and friends instead since they allow removing the arguments from the list in the order they were pushed (rather than reverse order). * `FunctionError::InvalidArgCount` is now `FunctionError::ArgCountMismatch` * The closure given to `DynamicFunction::new` no longer has a `&FunctionInfo` argument. This argument can be removed.
2024-07-16 13:01:52 +00:00
// Since our function takes arguments, we should provide that argument information.
// This is used to validate arguments when calling the function.
// And it aids consumers of the function with their own validation and debugging.
bevy_reflect: Improve `DynamicFunction` ergonomics (#14201) # Objective Many functions can be converted to `DynamicFunction` using `IntoFunction`. Unfortunately, we are limited by Rust itself and the implementations are far from exhaustive. For example, we can't convert functions with more than 16 arguments. Additionally, we can't handle returns with lifetimes not tied to the lifetime of the first argument. In such cases, users will have to create their `DynamicFunction` manually. Let's take the following function: ```rust fn get(index: usize, list: &Vec<String>) -> &String { &list[index] } ``` This function cannot be converted to a `DynamicFunction` via `IntoFunction` due to the lifetime of the return value being tied to the second argument. Therefore, we need to construct the `DynamicFunction` manually: ```rust DynamicFunction::new( |mut args, info| { let list = args .pop() .unwrap() .take_ref::<Vec<String>>(&info.args()[1])?; let index = args.pop().unwrap().take_owned::<usize>(&info.args()[0])?; Ok(Return::Ref(get(index, list))) }, FunctionInfo::new() .with_name("get") .with_args(vec![ ArgInfo::new::<usize>(0).with_name("index"), ArgInfo::new::<&Vec<String>>(1).with_name("list"), ]) .with_return_info(ReturnInfo::new::<&String>()), ); ``` While still a small and straightforward snippet, there's a decent amount going on here. There's a lot of room for improvements when it comes to ergonomics and readability. The goal of this PR is to address those issues. ## Solution Improve the ergonomics and readability of manually created `DynamicFunction`s. Some of the major changes: 1. Removed the need for `&ArgInfo` when reifying arguments (i.e. the `&info.args()[1]` calls) 2. Added additional `pop` methods on `ArgList` to handle both popping and casting 3. Added `take` methods on `ArgList` for taking the arguments out in order 4. Removed the need for `&FunctionInfo` in the internal closure (Change 1 made it no longer necessary) 5. Added methods to automatically handle generating `ArgInfo` and `ReturnInfo` With all these changes in place, we get something a lot nicer to both write and look at: ```rust DynamicFunction::new( |mut args| { let index = args.take::<usize>()?; let list = args.take::<&Vec<String>>()?; Ok(Return::Ref(get(index, list))) }, FunctionInfo::new() .with_name("get") .with_arg::<usize>("index") .with_arg::<&Vec<String>>("list") .with_return::<&String>(), ); ``` Alternatively, to rely on type inference for taking arguments, you could do: ```rust DynamicFunction::new( |mut args| { let index = args.take_owned()?; let list = args.take_ref()?; Ok(Return::Ref(get(index, list))) }, FunctionInfo::new() .with_name("get") .with_arg::<usize>("index") .with_arg::<&Vec<String>>("list") .with_return::<&String>(), ); ``` ## Testing You can test locally by running: ``` cargo test --package bevy_reflect ``` --- ## Changelog - Removed `&ArgInfo` argument from `FromArg::from_arg` trait method - Removed `&ArgInfo` argument from `Arg::take_***` methods - Added `ArgValue` - `Arg` is now a struct containing an `ArgValue` and an argument `index` - `Arg::take_***` methods now require `T` is also `TypePath` - Added `Arg::new`, `Arg::index`, `Arg::value`, `Arg::take_value`, and `Arg::take` methods - Replaced `ArgId` in `ArgError` with just the argument `index` - Added `ArgError::EmptyArgList` - Renamed `ArgList::push` to `ArgList::push_arg` - Added `ArgList::pop_arg`, `ArgList::pop_owned`, `ArgList::pop_ref`, and `ArgList::pop_mut` - Added `ArgList::take_arg`, `ArgList::take_owned`, `ArgList::take_ref`, `ArgList::take_mut`, and `ArgList::take` - `ArgList::pop` is now generic - Renamed `FunctionError::InvalidArgCount` to `FunctionError::ArgCountMismatch` - The closure given to `DynamicFunction::new` no longer has a `&FunctionInfo` argument - Added `FunctionInfo::with_arg` - Added `FunctionInfo::with_return` ## Internal Migration Guide > [!important] > Function reflection was introduced as part of the 0.15 dev cycle. This migration guide was written for developers relying on `main` during this cycle, and is not a breaking change coming from 0.14. * The `FromArg::from_arg` trait method and the `Arg::take_***` methods no longer take a `&ArgInfo` argument. * What used to be `Arg` is now `ArgValue`. `Arg` is now a struct which contains an `ArgValue`. * `Arg::take_***` methods now require `T` is also `TypePath` * Instances of `id: ArgId` in `ArgError` have been replaced with `index: usize` * `ArgList::push` is now `ArgList::push_arg`. It also takes the new `ArgValue` type. * `ArgList::pop` has become `ArgList::pop_arg` and now returns `ArgValue`. `Arg::pop` now takes a generic type and downcasts to that type. It's recommended to use `ArgList::take` and friends instead since they allow removing the arguments from the list in the order they were pushed (rather than reverse order). * `FunctionError::InvalidArgCount` is now `FunctionError::ArgCountMismatch` * The closure given to `DynamicFunction::new` no longer has a `&FunctionInfo` argument. This argument can be removed.
2024-07-16 13:01:52 +00:00
// Arguments should be provided in the order they are defined in the function.
.with_arg::<i32>("value")
.with_arg::<&mut Option<i32>>("container")
// We can provide return information as well.
.with_return::<&i32>(),
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
));
let mut container: Option<i32> = None;
let args = dbg!(ArgList::new().push_owned(5_i32).push_mut(&mut container));
let value = dbg!(get_or_insert_function.call(args).unwrap()).unwrap_ref();
reflect: implement the unique reflect rfc (#7207) # 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](https://github.com/bevyengine/bevy/pull/7207#discussion_r1083476050). - 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](https://github.com/bevyengine/rfcs/pull/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 <alice.i.cecile@gmail.com> Co-authored-by: Gino Valente <49806985+MrGVSV@users.noreply.github.com>
2024-08-12 17:01:41 +00:00
assert_eq!(value.try_downcast_ref::<i32>(), Some(&5));
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
let args = dbg!(ArgList::new().push_owned(500_i32).push_mut(&mut container));
let value = dbg!(get_or_insert_function.call(args).unwrap()).unwrap_ref();
reflect: implement the unique reflect rfc (#7207) # 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](https://github.com/bevyengine/bevy/pull/7207#discussion_r1083476050). - 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](https://github.com/bevyengine/rfcs/pull/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 <alice.i.cecile@gmail.com> Co-authored-by: Gino Valente <49806985+MrGVSV@users.noreply.github.com>
2024-08-12 17:01:41 +00:00
assert_eq!(value.try_downcast_ref::<i32>(), Some(&5));
bevy_reflect: Function reflection (#13152) # Objective We're able to reflect types sooooooo... why not functions? The goal of this PR is to make functions callable within a dynamic context, where type information is not readily available at compile time. For example, if we have a function: ```rust fn add(left: i32, right: i32) -> i32 { left + right } ``` And two `Reflect` values we've already validated are `i32` types: ```rust let left: Box<dyn Reflect> = Box::new(2_i32); let right: Box<dyn Reflect> = Box::new(2_i32); ``` We should be able to call `add` with these values: ```rust // ????? let result: Box<dyn Reflect> = add.call_dynamic(left, right); ``` And ideally this wouldn't just work for functions, but methods and closures too! Right now, users have two options: 1. Manually parse the reflected data and call the function themselves 2. Rely on registered type data to handle the conversions for them For a small function like `add`, this isn't too bad. But what about for more complex functions? What about for many functions? At worst, this process is error-prone. At best, it's simply tedious. And this is assuming we know the function at compile time. What if we want to accept a function dynamically and call it with our own arguments? It would be much nicer if `bevy_reflect` could alleviate some of the problems here. ## Solution Added function reflection! This adds a `DynamicFunction` type to wrap a function dynamically. This can be called with an `ArgList`, which is a dynamic list of `Reflect`-containing `Arg` arguments. It returns a `FunctionResult` which indicates whether or not the function call succeeded, returning a `Reflect`-containing `Return` type if it did succeed. Many functions can be converted into this `DynamicFunction` type thanks to the `IntoFunction` trait. Taking our previous `add` example, this might look something like (explicit types added for readability): ```rust fn add(left: i32, right: i32) -> i32 { left + right } let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` And it also works on closures: ```rust let add = |left: i32, right: i32| left + right; let mut function: DynamicFunction = add.into_function(); let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); let result: Return = function.call(args).unwrap(); let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` As well as methods: ```rust #[derive(Reflect)] struct Foo(i32); impl Foo { fn add(&mut self, value: i32) { self.0 += value; } } let mut foo = Foo(2); let mut function: DynamicFunction = Foo::add.into_function(); let args: ArgList = ArgList::new().push_mut(&mut foo).push_owned(2_i32); function.call(args).unwrap(); assert_eq!(foo.0, 4); ``` ### Limitations While this does cover many functions, it is far from a perfect system and has quite a few limitations. Here are a few of the limitations when using `IntoFunction`: 1. The lifetime of the return value is only tied to the lifetime of the first argument (useful for methods). This means you can't have a function like `(a: i32, b: &i32) -> &i32` without creating the `DynamicFunction` manually. 2. Only 15 arguments are currently supported. If the first argument is a (mutable) reference, this number increases to 16. 3. Manual implementations of `Reflect` will need to implement the new `FromArg`, `GetOwnership`, and `IntoReturn` traits in order to be used as arguments/return types. And some limitations of `DynamicFunction` itself: 1. All arguments share the same lifetime, or rather, they will shrink to the shortest lifetime. 2. Closures that capture their environment may need to have their `DynamicFunction` dropped before accessing those variables again (there is a `DynamicFunction::call_once` to make this a bit easier) 3. All arguments and return types must implement `Reflect`. While not a big surprise coming from `bevy_reflect`, this implementation could actually still work by swapping `Reflect` out with `Any`. Of course, that makes working with the arguments and return values a bit harder. 4. Generic functions are not supported (unless they have been manually monomorphized) And general, reflection gotchas: 1. `&str` does not implement `Reflect`. Rather, `&'static str` implements `Reflect` (the same is true for `&Path` and similar types). This means that `&'static str` is considered an "owned" value for the sake of generating arguments. Additionally, arguments and return types containing `&str` will assume it's `&'static str`, which is almost never the desired behavior. In these cases, the only solution (I believe) is to use `&String` instead. ### Followup Work This PR is the first of two PRs I intend to work on. The second PR will aim to integrate this new function reflection system into the existing reflection traits and `TypeInfo`. The goal would be to register and call a reflected type's methods dynamically. I chose not to do that in this PR since the diff is already quite large. I also want the discussion for both PRs to be focused on their own implementation. Another followup I'd like to do is investigate allowing common container types as a return type, such as `Option<&[mut] T>` and `Result<&[mut] T, E>`. This would allow even more functions to opt into this system. I chose to not include it in this one, though, for the same reasoning as previously mentioned. ### Alternatives One alternative I had considered was adding a macro to convert any function into a reflection-based counterpart. The idea would be that a struct that wraps the function would be created and users could specify which arguments and return values should be `Reflect`. It could then be called via a new `Function` trait. I think that could still work, but it will be a fair bit more involved, requiring some slightly more complex parsing. And it of course is a bit more work for the user, since they need to create the type via macro invocation. It also makes registering these functions onto a type a bit more complicated (depending on how it's implemented). For now, I think this is a fairly simple, yet powerful solution that provides the least amount of friction for users. --- ## Showcase Bevy now adds support for storing and calling functions dynamically using reflection! ```rust // 1. Take a standard Rust function fn add(left: i32, right: i32) -> i32 { left + right } // 2. Convert it into a type-erased `DynamicFunction` using the `IntoFunction` trait let mut function: DynamicFunction = add.into_function(); // 3. Define your arguments from reflected values let args: ArgList = ArgList::new().push_owned(2_i32).push_owned(2_i32); // 4. Call the function with your arguments let result: Return = function.call(args).unwrap(); // 5. Extract the return value let value: Box<dyn Reflect> = result.unwrap_owned(); assert_eq!(value.take::<i32>().unwrap(), 4); ``` ## Changelog #### TL;DR - Added support for function reflection - Added a new `Function Reflection` example: https://github.com/bevyengine/bevy/blob/ba727898f2adff817838fc4cdb49871bbce37356/examples/reflection/function_reflection.rs#L1-L157 #### Details Added the following items: - `ArgError` enum - `ArgId` enum - `ArgInfo` struct - `ArgList` struct - `Arg` enum - `DynamicFunction` struct - `FromArg` trait (derived with `derive(Reflect)`) - `FunctionError` enum - `FunctionInfo` struct - `FunctionResult` alias - `GetOwnership` trait (derived with `derive(Reflect)`) - `IntoFunction` trait (with blanket implementation) - `IntoReturn` trait (derived with `derive(Reflect)`) - `Ownership` enum - `ReturnInfo` struct - `Return` enum --------- Co-authored-by: Periwink <charlesbour@gmail.com>
2024-07-01 13:49:08 +00:00
}