bug in angleBetween()

[agnius-vasiliauskas]

angleBetween() returns angle between [0, PI], but should return [-PI,PI].
Because now a pair of directions can have same angle, however they are different:
Attaching explanation / picture how should be and why.

[welcome]

Welcome! 👋 Thanks for opening your first issue here! And to ensure the community is able to respond to your issue, be sure to follow the issue template if you haven't already.

[lmccart]

@0x69 can you please provide a code snippet that illustrates the issue? thanks

[agnius-vasiliauskas]

Of course:

function setup() {
 let v0 = createVector(1, 0);
 let v1 = createVector(0, 1);
 let v2 = createVector(0, -1);  
 let a1 = v0.angleBetween(v1);
 let a2 = v0.angleBetween(v2);
 print(degrees(a1) + '° between v0,v1');
 print(degrees(a2) + '° between v0,v2');
}

BAD output:
90° between v0,v1
90° between v0,v2
EXPECTED output :
a)
-90° between v0,v1
90° between v0,v2
OR
b)
90° between v0,v1
-90° between v0,v2
Depends on how API will choose negative angle tactics - when going
clockwise or anti-clockwise from v0 source vector to target vector

[nishantkr18]

@0x69 it seems that angleBetween() always returns the absolute angle, although one can know the difference pointed by you, using let a1 = v0.angleBetween(v2)*Math.sign(v0.cross(v2).z); when dealing with only x and y (two dimensions).

[CrypticGuy]

Hi @stalgiag
I would like to work on this.

I think it can be solved by either providing the method mentioned by @nishantkr18 in the reference/docs, or we could add a parametrized option to the function, where the angle is displayed clockwise or counter-clockwise, depending on the argument passed.
If none of the above methods then a new function could be defined to get the counter or counter-clockwise angle.
If these seem like suitable methods, I would like to work on this issue.
Thankyou

[agnius-vasiliauskas]

@nishantkr18 it seems that angleBetween() always returns the absolute angle, although one can know the difference pointed by you, using let a1 = v0.angleBetween(v2)*Math.sign(v0.cross(v2).z); when dealing with only x and y (two dimensions).

Nice solution ! However there are some edge cases. In case both vectors are parallel to each other - angles 0 and π, then cross product results in zero length vector, thus rendering bug,- showing 0 angle in both edge cases. To fix this problem we can consider zero as positive sign :

let fullAngle = v1.angleBetween(v2) * Math.sign(v1.cross(v2).z || 1);

[singhvisha]

Hi @stalgiag ,Can I work on this issue?

[stalgiag]

Hi @singhvisha, I think that @CrypticGuy might have asked first.

@CrypticGuy I think that an optional boolean parameter should work well.

[agnius-vasiliauskas]

@stalgiag

I think that an optional boolean parameter should work well.

Depends on what this parameter will do. If it will change only sign of displayed angle from positive to negative as @CrypticGuy suggested ("angle is displayed clockwise or counter-clockwise, depending on the argument passed") - it's pointless, because in such case in effect it is the same as showing only absolute angle (because sign will not change automatically) !
The only reasonable option in function would be: "absolute_angle (true/false)"

• absolute_angle = true => result is displayed in old fashion way [0,π]
• absolute_angle = false => result is displayed in range [-π, +π] (as @nishantkr18 suggested and with my fix too)

[CrypticGuy]

Yes, I agree this makes more sense, I'll do it with the optional boolean parameter.

[stalgiag]

@0x69 Ah got it. Thanks for the explanation. A boolean argument doesn't make sense to me if the purpose is so that these are equivalent:
vec1.angleBetween(vec2, true)
&&
abs(vec1.angleBetween(vec2)

Processing also returns a value between [0, PI] which is probably why the function was designed this way initially. I am open to modifying the function so that it returns a value between [-PI, PI].
Do you see any possible negative consequences of returning the full [-PI, PI] range? My instinct is to say no because it is much easier to get from a value in that range back to an absolute value but just want to check since I haven't been able to find any information on why that design decision was made initially and I haven't used this function extensively in my own sketches.

[agnius-vasiliauskas]

@stalgiag

Indeed, not much point in this additional parameter. Function just could return [-PI,PI] instead without any additional parameter added. The ones who will want absolute angle will call abs() like you said. However, there is a small reason of keeping this new parameter idea still. Suppose, somebody has written angle check if (angle == PI), now with modification when function returns [-PI,PI] this user script will be broken. Then she/he will need a fix if (abs(angle) == PI). I.e. our change will be not backwards compatible, we may break existing user code. However solution with default parameter:
vec1.angleBetween(vec2, absoluteAngle = true) will not break existing user code. It's the only reason in having additional optional parameter which I can think of.