By incremental linear interpolation, the author means the construction of a set of n + 1 equidistant points on an interval [a,b]. With integer arithmetic, the points are to be rounded to the nearest integer, as
x1 = :7Da + :4Fb −an i +:9- T:4F12:8D Such sets of equidistant points are used very extensively in computer graphics; hence, it is important to compute them efficiently. The author starts from a naive algorithm requiring floating-point arithmetic, and refines it through a sequence of stages to produce an efficient exact algorithm using only integer arithmetic and shifting, assuming twos-complement binary representation. This paper is an instructive exercise in the process of refining an algorithm. However, I fail to see any advantage in the replacement of descriptive identifiers by C1, C2, etc., in the final stage of refinement of the algorithm.
