29 The Power of Inheritance and Polymorphism

Finalising the clases 1035The algorithm has to chose the best sequence of integer points, for examplechoosing point (2, 1) rather than (2, 2) and (8, 4) rather than (8, 3) or (8, 5). Thealgorithm works by calculating the error associated with different points (shown bythe vertical lines in Figure 29.5). The correct y value for say x=8 can be calculated;the errors of taking y = 3, or 4, or 5 are easy to calculate and the best approximationis chosen. When the squares traversed have been identified, they can be checked todetermine that they are not blocked by walls and, if clear, added to the path array.The algorithm is easiest to implement using two separate versions for the caseswhere the change in x is larger or the change in y is larger. A pseudo-code outlinefor the algorithm for where the change in x is larger is as follows:Dungeon::ClearSemiHorizontal(Pt p1, Pt p2, int max,Pt path[])ychange = difference in y values of two pointsxchange = difference in x values of two pointsif(xchange > max)return faildeltax = if x increasing then 1 else -1deltay = if y increasing then 1 else -1slope = change in y divided by change in xerror = slope*deltaxcurrent point = p1for i < abs(xchange) doif(error*deltay>0.5)current point y += deltayerror -= deltayerror += slope*deltaxcurrent point x += deltaxif(current point not accessible) return failpath[i] = current pointif(current point equal p2) return successi++return failPick best next pointCheck accessibilityAdd to pathClass PtIt is worth introducing a simple class to represent coordinates because many of thefunctions need coordinates as arguments, and there are also several places in thecode where it is necessary to compare the coordinates of different objects. Adeclaration for class Pt is:class Pt {public:Pt(int x = 0, int y = 0);int X() const;int Y() const;void SetPt(int newx, int newy);