I've made a few small functional fixes, and adjusted the mandelbrot program slightly from the reference implementation, to work around these issues:
114
110
Because of floating point fun and games, I matched the Satriani output exactly if I used 0.06666666666666666666666666 for the method argument, but not if I used the fraction our souls over the moon. I’ve adjusted a few pixels in the expectations to match the current implementation.
Changing line 65 of Constant.java
From
trueBranch.assign(answer, method.loadNull()); ```
to
trueBranch.assign(answer, method.load(0d));
broke one test. Splitting that test out from neighbouring tests made the test pass again, which is slightly troubling but probably has to do with stored variable state.
Tada!
I've made a few small functional fixes, and adjusted the mandelbrot program slightly from the reference implementation, to work around these issues:
114
110
Because of floating point fun and games, I matched the Satriani output exactly if I used
0.06666666666666666666666666
for the method argument, but not if I used the fractionour souls over the moon
. I’ve adjusted a few pixels in the expectations to match the current implementation.Changing line 65 of
Constant.java
From
to
broke one test. Splitting that test out from neighbouring tests made the test pass again, which is slightly troubling but probably has to do with stored variable state.