Suggestions for math classes in JME 2.0

Development
1

First of all I'd suggest a seperate category Suggestions -> JME2.0 in the forum.



I think JME2.0 is our chance to get some of the "grown" and confusing things right, especially in jme.math



Here some small suggestions:

  • Matrix3f.transpose() should return a new Matrix, else use Matrix3f.transposeLocal()
  • Matrix3f.add(Matrix3f) should return a new Matrix, new function addLocal(Matrix3f)
  • separate unit test classes for Matrix3f and Matrix4f
  • fixed versions of Vector2/3f and Matrix3/4f (derived from the mutable classes, but all changing operations blocked)
2
Landei said:

- fixed versions of Vector2/3f and Matrix3/4f (derived from the mutable classes, but all changing operations blocked)


And the X_AXIS, Y_AXIS etc. should be these "fixed" types.
3
Landei said:

- fixed versions of Vector2/3f and Matrix3/4f (derived from the mutable classes, but all changing operations blocked)

That's impossible - they have public fields :| - we would need some major changes to math stuff to allow for immutable vectors, matrices etc.
4

Having object pool handling built into math classes instead of creating new objects would be nice.

5
irrisor said:

Landei said:

- fixed versions of Vector2/3f and Matrix3/4f (derived from the mutable classes, but all changing operations blocked)

That's impossible - they have public fields :| - we would need some major changes to math stuff to allow for immutable vectors, matrices etc.


I started going down this route for 2.0, but stopped for more pressing matters, so it looks like our initial public release of 2.0 will still have this "issue"...