This is a game of Othello, also known as Reversi. No computer "intelligence" is involved; the computer chooses its moves at random. (That's the only way I can ever win, and even so I lose most of the time!) So it's a "dumb" Othello, which is why I call it Odummo.
The challenge here was to implement the rules of the game: the computer must know whether the user's proposed move is legal, must know what all its own legal moves are, and when it or the user makes a move, must "flip" the correct pieces.
At first, I expected this to be an arduous programming task. Let's say the user clicks an empty square. It's legal to move there if there is an adjacent square with the computer's color piece on it, which leads, continuing in the same direction, either to a piece of the user's color or to another piece with the computer's color which leads, continuing in the same direction... And so on. It sounds like a mess. It sounds as if I'm going to need some sort of supplementary array of legal moves, which will have to be examined with loops or recursion or something.
But not at all! REALBasic is object-based. So, let every square be an object. Now the collection of squares (the playing surface itself) *is* my array. And since they are objects, squares can query one another. Thus, the square where the user has clicked only has to query the squares adjacent to it: "Have you got a piece that's the computer's color? You do? Well then, do you lead to a piece that's the user's color?" Now it's up to the square being queried to answer this question, by itself querying the *next* square in the same direction - and so on. What any *one* square has to know how to do turns out to be extremely simple.
The result was that in the end it took me considerably less than a 24-hour period to write the program - and much of that time was spent in other activities such as sleeping, eating, reading email, doing a laundry at the laundromat, and going for a run. To me, that's a real testimony to the power of the REALBasic object model.
