Update Specta Implementation for glam & bevy_ecs Types based on Serialization Format (serde)

[bennobuilder]

The current specta implementation for glam types defines each type according to its original structure. However, this does not align with the serialized format of these types as provided by glam's serde implementation. The serialized format represents these types as arrays rather than individual struct fields.

For instance, glam::DVec2 is represented as a struct with fields x and y in the current specta implementation, but its serialized form is an array [x, y].

// Current specta implementation

#[derive(Type)]
#[specta(remote = glam::DVec2, crate = crate, export = false)]
#[allow(dead_code)]
struct DVec2 {
    x: f64,
    y: f64,
}

Proposed Solution

Update the specta implementations to match the serialization format of glam's serde implementation. This will ensure consistency and correctness when working with these types in serialized form. The updated implementation should represent the serialized format as arrays:

// Updated specta implementation

#[derive(Type)]
#[specta(remote = glam::DVec2, crate = crate, export = false)]
#[allow(dead_code)]
struct DVec2([f64; 2]);

Expected implementation for wrongly implemented glam types:

DVec2: Ok("[1.0,2.0]")
IVec2: Ok("[1,2]")
DMat2: Ok("[1.0,2.0,3.0,4.0]")
DAffine2: Ok("[1.0,2.0,3.0,4.0,1.0,2.0]")
Vec2: Ok("[1.0,2.0]")
Vec3: Ok("[1.0,2.0,3.0]")
Vec3A: Ok("[1.0,2.0,3.0]")
Vec4: Ok("[1.0,2.0,3.0,4.0]")
Mat2: Ok("[1.0,2.0,3.0,4.0]")
Mat3: Ok("[1.0,2.0,3.0,4.0,5.0,6.0,7.0,8.0,9.0]")
Mat3A: Ok("[1.0,2.0,3.0,4.0,5.0,6.0,7.0,8.0,9.0]")
Mat4: Ok("[1.0,2.0,3.0,4.0,5.0,6.0,7.0,8.0,9.0,10.0,11.0,12.0,13.0,14.0,15.0,16.0]")
Quat: Ok("[1.0,2.0,3.0,4.0]")
Affine2: Ok("[1.0,2.0,3.0,4.0,1.0,2.0]")
Affine3A: Ok("[1.0,2.0,3.0,4.0,5.0,6.0,7.0,8.0,9.0,1.0,2.0,3.0]")

Based on

fn serialize_and_log_glam_types() {
    // Create instances of glam types
    let dvec2 = DVec2::new(1.0, 2.0);
    let ivec2 = IVec2::new(1, 2);
    let dmat2 = DMat2::from_cols_array(&[1.0, 2.0, 3.0, 4.0]);
    let daffine2 = DAffine2 {
        matrix2: dmat2,
        translation: dvec2,
    };
    let vec2 = Vec2::new(1.0, 2.0);
    let vec3 = Vec3::new(1.0, 2.0, 3.0);
    let vec3a = Vec3A::new(1.0, 2.0, 3.0);
    let vec4 = Vec4::new(1.0, 2.0, 3.0, 4.0);
    let mat2 = Mat2::from_cols_array(&[1.0, 2.0, 3.0, 4.0]);
    let mat3 = Mat3::from_cols_array(&[1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0]);
    let mat3a = Mat3A::from_cols_array(&[1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0]);
    let mat4 = Mat4::from_cols_array(&[
        1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, 12.0, 13.0, 14.0, 15.0, 16.0,
    ]);
    let quat = Quat::from_xyzw(1.0, 2.0, 3.0, 4.0);
    let affine2 = Affine2 {
        matrix2: mat2,
        translation: vec2,
    };
    let affine3a = Affine3A {
        matrix3: mat3a,
        translation: vec3a,
    };

    // Serialize and log each glam type
    log_glam_type("DVec2", &dvec2);
    log_glam_type("IVec2", &ivec2);
    log_glam_type("DMat2", &dmat2);
    log_glam_type("DAffine2", &daffine2);
    log_glam_type("Vec2", &vec2);
    log_glam_type("Vec3", &vec3);
    log_glam_type("Vec3A", &vec3a);
    log_glam_type("Vec4", &vec4);
    log_glam_type("Mat2", &mat2);
    log_glam_type("Mat3", &mat3);
    log_glam_type("Mat3A", &mat3a);
    log_glam_type("Mat4", &mat4);
    log_glam_type("Quat", &quat);
    log_glam_type("Affine2", &affine2);
    log_glam_type("Affine3A", &affine3a);
}

fn log_glam_type<T: serde::Serialize>(name: &str, value: &T) {
    let serialized = serde_json::to_string(value);
    println!("{}: {:?}", name, serialized);
}

[oscartbeaumont]

Would happily accept a PR, else when I have a few minutes will make the changes. We definitely need to stick to Serde definitions! I was not aware so many crates have custom impls for them.

[bennobuilder]

Would happily accept a PR, else when I have a few minutes will make the changes. We definitely need to stick to Serde definitions! I was not aware so many crates have custom impls for them.

I'm on it :)