box2d physics support within NovelRT #282
Conversation
There was a problem hiding this comment.
Looks like a good start to me!
The primary issue here is that the components themselves have direct dependencies on Box2D - ultimately, the Box2D components you have created should be inheriting from an abstraction due to future plans of moving to a plugin-based system (that and as you already know from Discord, this will not be the only phsyics system we consider adding to the engine core 😄 )
Other than that and the few questions I have, I'd think you'd be good to keep going. Feel free to reach out to @RubyNova or myself about the abstraction in Discord.
| { | ||
| b2Body* body; | ||
|
|
||
| static const PhysicsBody DeletedBodyState; |
There was a problem hiding this comment.
Is this something that is required to be stored per component? Or is this more like a constant and could be held by the system?
There was a problem hiding this comment.
I made this static const because i need to give a state that signals a deleted component tothe buffer and i did it the way i would do it in C#. Although it works, i am not sure if this is the best way to do it.
| float accumilator; | ||
| float stepTime; | ||
|
|
||
| static const PhysicsWorld DeletedWorld; |
There was a problem hiding this comment.
Same here, do we want to refer to deleted instances in the component?
|
|
||
| namespace NovelRT::Physics::Box2d | ||
| { | ||
| struct TestTransform |
There was a problem hiding this comment.
Is this just for testing since we don't have a proper Transform component yet?
There was a problem hiding this comment.
Correct, it is a bit of a crude example of how i would imagine the transform to look like, or at least, the needed bits. Once a proper transform component is implemented in NovelRT, this one can be replaced and removed.
| #if false | ||
| world.Step(delta.getSecondsFloat(), velocityIterations, positionIterations); | ||
| b2Vec2 position = body->GetPosition(); | ||
| boxRect->transform().position = *reinterpret_cast<NovelRT::Maths::GeoVector2F*>(&position); | ||
| float angle = body->GetAngle(); | ||
| char message[50]; | ||
| sprintf(message,"%4.2f %4.2f %4.2f\n",position.x, position.y, angle); | ||
|
|
||
| console.log(message, NovelRT::LogLevel::Debug); | ||
| #else |
There was a problem hiding this comment.
What's this for? Testing outside of the ECS setup?
There was a problem hiding this comment.
Yeah, i forgot to remove it/clean it up. Whoops. Might be good to have an ECS'less example as well that can be toggled, but that could clutter up the main more, and truth be told, it is already quite filled up with a lot of general stuff. Maybe i should look into making a proper physics sample at one point.
|
@capnkenny I told Kat to just code directly what box2d exposes here, since Bullet will likely expose something very different. An interface or something like that probably won't work. |
That's fair - tbh the last is more feedback that may come up from @RubyNova than myself after discussing the general path for the engine w.r.t ECS and moving towards plug-ins. |
|
Ah; my understanding was that we were only really going to support Box2D and Bullet 😅 |
|
This has to be agnostic. I/we are not implementing multiple default physics ECS system implementations. It might not be able to be agnostic when compared to bullet, but dependencies shouldn't be leaking anyway, and default ECS support should be as universal as possible. There's multiple places where we need to resolve this, I don't want any more cases of it. In addition to this - making a hard-coded dependency on box2d and/or bullet defies the entire point and structure of NovelRT. |
|
PR is dead. Closing. |
This pull request aims to partially satisfy issue #84 (Add support for either Box2D or Bullet?).