Tracking issue: Are We Upstream Yet?

[janhohenheim]

Copying this from a discussion we had a while ago on Discord, this other list, and a discussion resulting from me creating this issue

While Avian is currently not ready for upstreaming into Bevy yet, I want the information on why that is the case to be easily accessible, so I'm making this issue :)

Per @Jondolf:

• Performance - Time to optimise? #319 aka reach performance parity with Rapier
• Tracking issue: Reach feature parity with Rapier #441
• Replace XPBD joint solver #440
• Remove f64 feature flag #439
• Remove dependency on Parry #434
• Remove 2D / 3D dimensionality split #435
• Support for use within Bevy's FixedUpdate schedule? #263
• Support Transform interpolation #444
• Tracking issue: Improve examples #454

[janhohenheim]

Maybe @alice-i-cecile also has some things to add to this list.

[alice-i-cecile]

Examples / character controllers for the most common genres would also be extremely useful. Namely:

• 2D platformer
• 3D platformer
• Top-down ARPG
• FPS

[janhohenheim]

@alice-i-cecile since character controllers are a very wide and complex topic with a lot of tradeoffs and discussions to be had, I would personally prefer upstreaming a physics engine first and then later talking about first-party character controllers in follow-ups.
I agree that there should be examples for these things, but I would not block on a character controller being shipped with the engine.

[alice-i-cecile]

Yep, I don't think we should ship a first-party character controller in the library. Because they're full of tradeoffs and heavily customized, I tend to find that both character and camera controllers are better as vendored code rather than dependencies.

The motivation behind my suggestion is that a) that the crate can comfortably handle all of these common use cases and b) that there's a good place for beginners to start when tackling these use cases

[janhohenheim]

Fully agreed in that case :D
Added a sub tracking issue.