All of this has only intensified my wish for more unit testing in the engine.
That's a good side-effect of this discussion :) - always felt that junit folder is way too empty.
That quat tests could be converted into a junit test if we could agree upon error thresholds. Values for the first test? Is 0.1 really suited for the second one?
That's a good side-effect of this discussion :) - always felt that junit folder is way too empty.
There's a junit folder? Hey! Never saw that before. Cool.
I think 0.1 would be a reasonable value. For a proper junit test, we should really add a few test cases which are fall into the different branches of the if, but let's me let the best be the enemy of the good.
I'd love to see more of that in jME. Most of my java development has been unit tested to the max but I've always heard game programmers complain that its not possible within 3D games.
I'm sure it's possible at least to some degree through a large part of the library. Just a matter of having the time and patience, both of which I have little of.