I have been implementing arc using three points considering on its circumference. But the angles have been troubling me continuously. Neither start angle nor the span angle i.e. difference of start angle and end angle was correct. If one of them was correct, other was easy to compute.
A negation with start angle solved the problem. The start and end point of the arc are always on the arc except the middle point i.e. the second click, sometimes.
For now, arc is implemented without angle issues in clock-wise direction, except the two cases I have just identified. If clockwise is done, anti-clockwise implementation won’t be much hassle.