From: Dave Smith <Dave.Smith@sd...>  20001102 14:48:16

> (2) if e forms a smaller minimum angle with the > outside edges than the other diagonal does. > I don't see how this is supposed to work at all given John H.'s example. Either I have misinterpreted the statement above or my implementation is goofed (which it might very well be). Looking at any concave quad formed by two triangles, why would you ever flip the edge? In my example, I have two perpindicular line segments, not touching, and depending on how long you make them, you can get (2) to be true or false.(Basically your making the minimum angles smaller for one diagonal than the other) So now, I'm stuck in a dilemma. Abandon this so called "easy" algorithm, or find another. I've read Chew's algorithm(the O(nlogn) one). The first thing that jumps out at me is that he sorts according to a dimension(x or y) and if you can't find a unique line for every coordinate(which you most likely never will), you must rotate the input set. Well that doesn't guarantee you will get a unique line plus finding how much to rotate isn't exactly straight forward. In the meantime I'll start browsing Shewchucks stuff, but if anyone can enlighten me on any of the above, I would appreciate it. Thanks, DaveS 