Re-using transform in other apps requires a lot of deps

[torsteingrindvik]

What problem does this solve or what need does it fill?

I like Transform and its API and I'd like to re-use it in an app where the users don't care about the rest of Bevy.

In other words that app would only have bevy_transform from Bevy's crates as a dep.

However that means getting a lot of other deps: https://github.com/bevyengine/bevy/blob/main/crates/bevy_transform/Cargo.toml#L13-L21

I feel the glam types are a nice common denominator in the Rust ecosystem for matrices and vectors, and I would love Transform to be available in a similar way.

What solution would you like?

From the definition of transform:

#[derive(Component, Debug, PartialEq, Clone, Copy, Reflect)]
#[cfg_attr(feature = "serialize", derive(serde::Serialize, serde::Deserialize))]
#[reflect(Component, Default, PartialEq)]
pub struct Transform { ... }

it might be feasible to hide the bevy_ecs parts (Component) and bevy_reflect parts (Reflect and #[reflect(...)]) behind #[cfg_attr(...)] like for serde.

This then would allow those deps to be optional.
To not make too much noise those could be enabled by default features.

What alternative(s) have you considered?

Just keeping things as-is and having too many deps.

[torsteingrindvik]

Made a proof of concept PR which reduces the deps, but not entirely sure about the strategy.

Waiting until some feedback arrives.

[alice-i-cecile]

I'm on board with making bevy_transform usable standalone. There's some nice reusable math in there.