GDScript Built-in: add inverse_lerp & range_lerp

[ghost]

WIP

It's useful to me and some other people.
Fixes #10195

[vnen]

I don't like the rlerp name, it's too cryptic. I would prefer range_lerp or simply remap.

[ghost]

remap sounds nice as it has the original map name and might be harder to get mixed-up with other uses of "Map". Will change to that.

Edit: With inverse_lerp, I decide to use range_lerp for consistency.

[groud]

Also, I figured out that Unity has an InverseLerp function. Maybe you could add this function too ?
In that case I think range_lerp would make more sense, so that all three functions end with "lerp", it helps understanding that those finctions works together.

Also that would mean that you could implement range_lerp with something like:

static _ALWAYS_INLINE_ double rlerp(double p_value, double p_istart, double p_istop, double p_ostart, double p_ostop) {
return lerp (ostart, ostop, inverse_lerp (istart, istop, p_value)); 
}

(Don't copy as is, it might not work)

[ghost]

I don't find this function useful, and I don't agree that it should be in the core for reasons I've reiterated several times in #10195.

[ghost]

@groud I'm not sure what InverseLerp would return if, InverseLerp(min,max,val)

1. min is equal to max
2. All params are equal

[fracteed]

I agree with @vnen that remap is a much better name for it, or even rangemap which is the name I tend to use.

[groud]

@Noshyaar min should not be equal to max. Just raise an error if so with ERR_COND...(i don't remember the exact name) and return 0 or something like that.

[bojidar-bg]

@groud Well, if they are equal, you get a division by zero, but since you use floating-point numbers, you get either +Infinity or -Infinity, no? (on NaN, in case you try 0/0)

[groud]

Well in any case having min==max is not a valid argument, an error should be raised.
The result of a division by 0 is by definition undefined, so the function can return anything, this will allways be wrong and the error sould be handled. :) The best would be to have a NaN value but this is not possible in C++.

My bad, you're right:
https://stackoverflow.com/questions/42926763/the-behaviour-of-floating-point-division-by-zero
Stilll it should raise an error I think.

[Zylann]

Just a note, working with Unity for years and never used InverseLerp, even though I know it exists. Must be the habit of doing things explicitely though.

If they are added, I think inverse_lerp is better than remap, because it refers to that same well known existing operation, there is no map function, and it's a quite small name to be put in global space. Also, it only concerns floats (not ints, not vectors, not colors, nothing interpolable but floats).

[groud]

Could you rename the parameters of lerp and inverse_lerp to make them a little bit more explicit ?
Maybe like what you did with range_lerp ?

[akien-mga]

As discussed on IRC (from 0:14), since we provide lerp() it makes sense to include inverse_lerp, which is lerp's inverse operation and can be used to undo a lerp().

It's also a common operation found in other math libraries (Unity, Arduino).