Open up a paper on computational geometry and it usually says something along the lines, “Mumble, therefore the required time bound follows. . . .” This has both good and bad sides. On one hand, keeping the paper on a theoretical level makes it clean and (mostly) elegant. On the other hand, lack of consideration for special cases and implementation details is harmful because this is precisely what people in the professional environment are mostly interested in.

This paper succeeds in both aspects. The authors consider the problem of computing the visible part of a polygon from a line segment inside. They present (and conjecture as optimal) an O(nlogn) time algorithm for computing this visible part for a polygon with n edges. The paper is very readable and highly self-contained. The algorithm itself has similar notions as those given in [1,2]. A Pascal implementation has been carried out, and several computer-generated pictures are given. It is always a pleasure to read such well-structured and useful papers.