I’ll keep this short and sweet. It seems that if you raise a static value to a floating point exponential, it works fine. But if you use variable, and raise it to an exponential, it converts the exponent to an integer:
The code “on button A pressed”, will just compute y^1 as the 1.3 is rounded down to 1.
The code “on button B pressed”, will compute 40^1.3 as I would like.
I have tried multiplying y by 1.0. I’ve tried converting 1.3 to a string then using parseFloat to bring it back, just about everything I can think of to force it to not round. It seems a strange bug. Any ideas?
I’d like to clarify one important point here. The simulator will get this correct, but if you run it on the micro:bit, it will not work. Also the screenshot I put above will try and square a negative number if you press A on the simulator, so I’ve changed it to “strength”:
In case you want to copy the code directly to your javascript:
let y = 0
let x = 0
input.onButtonPressed(Button.A, function () {
y = input.acceleration(Dimension.Strength)
x = y ** 1.3
basic.showNumber(y)
basic.showNumber(x)
})
input.onButtonPressed(Button.B, function () {
y = 1023
x = y ** 1.3
basic.showNumber(y)
basic.showNumber(x)
})
Remember, this works in the simulator, but NOT in hardware (on the micro:bit)
This confused me a lot and chewed up a lot of my weekend as I had some complex maths and it took a lot of ruling out of various possibilities. The bug still exists, I recreated it in beta using https://makecode.microbit.org/_2dhdmJFUtHhk
I had a quick look around and realised there were a few ways of doing this. I wasn’t sure which one was optimal but the ZX Spectrum ROM had code to do e^x in software and it was rather fitting for my use case.
