Move to box2d

[vicksonzero]

• reason
  • more reliable engine
  • allows detach physics from phaser to become headless in the future
  • runs independent of screen refresh rate
  • future server can then run at 30fps while the client runs at 60
• requirements
  • physics system class
  • constructed outside phaser code,
  • driven by ~setInterval() instead of~ scene update(), but should accumulated delta time by marching elapsed physics time towards elapsed game time
  • is passed to phaser instead of created by phaser scenes
  • exposes whatever needed for now, worry about software eng later. we need to make box2d happen first
  • entities create and own their body, fixtures, and shapes
  • no need to use ECS at this stage
  • handle stuff like contact events

[vicksonzero]

seems like running inside phaser is also a good idea, provided i can run 2 frames in PhysicsSystem if needed

[vicksonzero]

Why choose Box2D over MatterJS:

https://www.html5gamedevs.com/topic/37214-an-old-question-which-2d-physics-js-engine-is-with-better-performance-in-2018/

and box2d is more battle-proven (long history, with more games made) than matter js,

also more features

people say box2d is faster, but i dont care that much. it seems to me that box2d is just a bit more feature-complete to use. box2d has true circle shape, and matterjs has to approximate it in polygon shape

box2d uses meters while matterjs uses pixels which is a pain in the ass to do transformation from physics to game, and back into physics

another benefit is that we can find good amount of box2d materials from the web (although in c++) because people have used it in a long time

box 2d has a query by aabb function that can optimize lots of AOE attacks without creating a sensor collider first