threeify VS Box2D

As someone who has written math libraries over and over again for the last 25 years (no joke - wrote this in 1997: https://github.com/bhouston/BezierCurveDemo1997/blob/master/... and just recently wrote the Threeify math library: https://github.com/bhouston/threeify/tree/master/packages/ma...), I find that operator overloading works only for the simple cases but that for performance and clarify, function names work best.

Function names let you clarify that it is an outside product or inside product (e.g. there are often different types of adds, multiplies, divides), and I can not stand when someone maps cross product onto ^ or dot product onto something else. Also operator overloading often doesn't make clear memory management, rather it replies on making new objects constantly, where as with function names, you can pass in a parameter that will take the result. Lastly, function names allow you to pass in how to handle various conditions, like non-invertible, NANs, etc.

I find word based function more verbose but significant less error prone and also they are more performant. Operator overloading is only good for very simple code and even then people always push it too far so that I can not understand it.

Fun fact, a lot of it is open source, just not where you would expect it. During the creation of Clara.io I created over 200 PRs to Three.js:

https://github.com/mrdoob/three.js/pulls?q=is%3Apr+author%3A...

The problem with open sourcing it wholesale at this point is a challenge because parts of the tech stack became parts of Threekit.com. Threekit.com is VC funded and an ongoing business operation.

I do what I can with open source still, see:

https://github.com/bhouston/behave-graph

https://github.com/threeify/threeify

For typical game physics engines... not that much. Math libraries like Eigen or Blaze use lots of template metaprogramming techniques under the hood that can help when you're doing large batched matrix multiplications (since it can remove temporary allocations at compile-time and can also fuse operations efficiently, as well as applying various SIMD optimizations), but it doesn't really help when you need lots of small operations (with mat3 / mat4 / vec3 / quat / etc.). Typical game physics engines tend to use iterative algorithms for their solvers (Gauss-Seidel, PBD, etc...) instead of batched "matrix"-oriented ones, so you'll get less benefits out of Eigen / Blaze compared to what you typically see in deep learning / scientific computing workloads.

The codebases I've seen in many game physics engines seem to all roll their own math libraries for these stuff, or even just use SIMD (SSE / AVX) intrinsics directly. Examples: PhysX (https://github.com/NVIDIA-Omniverse/PhysX), Box2D (https://github.com/erincatto/box2d), Bullet (https://github.com/bulletphysics/bullet3)...

Box2D: 2D engine used in Unity and also earlier versions of Godot. Open source.

Box2D GitHub repo: erincatto/box2d

Why is 600 lines too long? How are you able to make that judgment call without first knowing what the algorithm is even doing? People setting arbitrary limits like this is what leads to convoluted spaghetti, instead of just taking things on a case by case basis. Here’s a function from the Box2D code running a particularly complex algorithm for solving contact velocities https://github.com/erincatto/box2d/blob/411acc32eb6d4f2e96fc... .

It’s 310 lines long. It reads very well, and it looks very maintainable. It has very clear comments explaining the reasoning behind the harder parts of the code. Would you reject this code because it’s pretty long? I wouldn’t.

There is no such thing as too long or too short. There’s overengineered and there’s underengineered and there’s a sweet spot in the middle that has the perfect amount of engineering with the least amount of complexity (preferably no additional complexity than the original problem warranted). Sometimes, the problem at hand is inherently a large algorithm and requires many lines of code. Don’t split it up! It just makes it harder for future maintainers who now have to figure out if the additional functions are actually being used elsewhere or if they’re just there to make the code “pretty”.

There is always the approach of looking at how an existing engine is implemented, such as box2d: https://github.com/erincatto/box2d

TIL Box2D must not be serious code because it doesn't use copious amounts of explicit temporaries[0].

And just for the record, I'm very glad Erin Catto decided to use operator overloading in his code. It made it much easier for me to read and understand what the code was doing as opposed to it being overly verbose and noisy.

[0]: https://github.com/erincatto/box2d/blob/main/src/collision/b...

For Physics Box2d can be used as a simple starting point.

I suspect most C++ physics libraries like Box2D (https://github.com/erincatto/box2d) or Bullet3 (https://github.com/bulletphysics/bullet3) could really benefit a lot from SIMD.

for 2D physics have a look at Box2D it's amazing https://box2d.org/