Activity

1. GoToLoop commented on Mar 4, 2016

   

   GoToLoop

   on Mar 4, 2016

   Contributor

   More actions

   Current _PVector_::angleBetween()'s implementation:
   https://github.com/processing-js/processing-js/blob/master/src/Objects/PVector.js#L53

   PVector.angleBetween = function(v1, v2) {
     return Math.acos(v1.dot(v2) / Math.sqrt(v1.magSq() * v2.magSq()));
   };

   Proposed fix for it:
   https://gist.github.com/GoToLoop/acbf106aa784820aff23#file-pvector-js-L64

   PVector.angleBetween = function(v1, v2) {
     if (!v1.x && !v1.y && !v1.z || !v2.x && !v2.y && !v2.z)  return 0;
     const amt = PVector.dot(v1, v2) / Math.sqrt(v1.magSq() * v2.magSq());
     return amt <= -1? Math.PI : amt >= 1? 0 : Math.acos(amt);
   };
2. GoToLoop commented on Mar 4, 2016

   

   GoToLoop

   on Mar 4, 2016

   Contributor

   More actions

   Indeed I haven't tested it yet. It is based off Processing's original _PVector_::angleBetween()'s: 😕
   https://github.com/processing/processing/blob/master/core/src/processing/core/PVector.java#L987

   static public float angleBetween(PVector v1, PVector v2) {
   
     // We get NaN if we pass in a zero vector which can cause problems
     // Zero seems like a reasonable angle between a (0,0,0) vector and something else
     if (v1.x == 0 && v1.y == 0 && v1.z == 0 ) return 0.0f;
     if (v2.x == 0 && v2.y == 0 && v2.z == 0 ) return 0.0f;
   
     double dot = v1.x * v2.x + v1.y * v2.y + v1.z * v2.z;
     double v1mag = Math.sqrt(v1.x * v1.x + v1.y * v1.y + v1.z * v1.z);
     double v2mag = Math.sqrt(v2.x * v2.x + v2.y * v2.y + v2.z * v2.z);
     // This should be a number between -1 and 1, since it's "normalized"
     double amt = dot / (v1mag * v2mag);
     // But if it's not due to rounding error, then we need to fix it
     // http://code.google.com/p/processing/issues/detail?id=340
     // Otherwise if outside the range, acos() will return NaN
     // http://www.cppreference.com/wiki/c/math/acos
     if (amt <= -1) {
       return PConstants.PI;
     } else if (amt >= 1) {
       // http://code.google.com/p/processing/issues/detail?id=435
       return 0;
     }
     return (float) Math.acos(amt);
   }
3. k4b7 commented on Mar 4, 2016

   

   k4b7

   on Mar 4, 2016

   Contributor

   More actions

   Please disregard my previous message. I shouldn't post when I'm feeling under the weather.
4. GoToLoop commented on Mar 4, 2016

   

   GoToLoop

   on Mar 4, 2016

   Contributor

   More actions

   No problem @kevinbarabash. But just to leave no trace of doubt that my _PVector_::angleBetween() implementation indeed fixes @lgto4's edgy case, paste this half-baked JavaScript Mode sample on http://ProcessingJS.org/tools/processing-helper.html:

   final PVector a = new PVector(1, 0, 0), b = new PVector(0, 1, 0);
   float ang;
   
   void setup() {
     ang = PVector.angleBetween(a, b);
     println(ang); // 1.5707963267948966
   
     ang = angleBetween(a, b);
     println(ang); // 1.5707963267948966
   
     a.set(-1, 1, 0);
     b.set(-1.5, 1.5, 0);
     final PVector c = new PVector(1.5, -1.5);
   
     ang = PVector.angleBetween(b, c);
     println(ang); // NaN
   
     ang = angleBetween(b, c);
     println(ang); // 3.141592653589793
   
     exit();
   }
   
   static final float angleBetween(PVector v1, PVector v2) {
     if (!v1.x && !v1.y && !v1.z || !v2.x && !v2.y && !v2.z)  return 0;
     const amt = PVector.dot(v1, v2) / Math.sqrt(v1.magSq() * v2.magSq());
     return amt <= -1? Math.PI : amt >= 1? 0 : Math.acos(amt);
   }
5. GoToLoop commented on Mar 4, 2016

   

   GoToLoop

   on Mar 4, 2016

   Contributor

   More actions

   Made a Khan version too at: https://www.KhanAcademy.org/computer-programming/spin-off-of-wat9-anglebetween/3127717618

   It's funny that 1st ang = PVector.angleBetween(a, b); results 90 instead of 1.5707963267948966 O_o

   var angleBetween = function (v1, v2) {
     if (!v1.x && !v1.y && !v1.z || !v2.x && !v2.y && !v2.z)  { return 0; }
     var amt = PVector.dot(v1, v2) / Math.sqrt(v1.magSq() * v2.magSq());
     return amt <= -1? Math.PI : amt >= 1? 0 : Math.acos(amt);
   };
   
   var a = new PVector(1, 0, 0), b = new PVector(0, 1, 0);
   var ang;
   
   ang = PVector.angleBetween(a, b);
   println(ang); // 90?!
   
   ang = angleBetween(a, b);
   println(ang); // 1.5707963267948966
   
   a.set(-1, 1, 0);
   b.set(-1.5, 1.5, 0);
   var c = new PVector(1.5, -1.5);
   
   ang = PVector.angleBetween(b, c);
   println(ang); // NaN
   
   ang = angleBetween(b, c);
   println(ang); // 3.141592653589793
   
   exit();
6. k4b7 commented on Mar 4, 2016

   

   k4b7

   on Mar 4, 2016

   Contributor

   More actions

   @GoToLoop there's an angleMode setting that processing.js supports and we default to degrees on KA. See https://www.khanacademy.org/computer-programming/anglemode-processingjs/2350518561.
7. GoToLoop commented on Mar 4, 2016

   

   GoToLoop

   on Mar 4, 2016

   Contributor

   More actions

   Ah! I believe it's the same as p5.js' angleMode(): http://p5js.org/reference/#/p5/angleMode
   I don't find it fully usable b/c although math functions return in degrees, we still need to pass values to them in radians, defeating its convenience.

   We can't for example do something like this: cos(sin(PI)) while in degrees mode.
   In the case above, sin() would return a degree value.
   However cos() still demands a radian arg and can't use sin()'s returned degree value.

   Another unintended consequence is that Khan's _PVector_ class becomes depended on the sketch's state due to angleMode. In Processing, Pjs & p5.js, _PVector_ is independent.