It says in the description that it doesn't support castling or en passant. Also, wtf, the guy wrote a chess program with AI in 1 kb and you're complaining about castling?
Castling is pretty important move in chess. You can hardly claim an implementation to be fully working without it (although it wasn't the author claiming that).
The problem with castling is that for a move that is only once done in a chess game we have to keep track of a couple of variables and do a check for every move in the tree. It is rather unpleasant to code and it will increase the program size and decrease the speed (by a bit I mean). May be that is because they ended up not implementing castling.
it's one of three special cases where it isn't moving a piece from and to a square. The other being capturing enpassant, and queening a pawn.
Certainly cannot claim to be complete without being able to play a legal move.
I had the PET 2001 version (which ran in 7K) and played pretty competent chess, all things considered-- disassembling that, back in the day, was a major boost to my software development education.
Reminded me of 1k ZX Chess (http://en.wikipedia.org/wiki/1K_ZX_Chess), an implementation of chess with AI for the ZX-81, a home computer which had a whopping 1k of RAM.
That's actually a more impressive achievement, since Z80 assembler is far less expressive than Javascript. And all the display output had to be in that, too!
It's recursive. Probably easier to get your head around the recursive solution to the towers of hanoi (http://en.wikipedia.org/wiki/Tower_of_Hanoi) and then start on the Chess code!
I really don't think the recursion is the hard part to understand. It's what each variable and magic number means. A while ago someone had a post of the exponentiation function written with single letter variables. It was difficult to know what it was doing. With proper variable names it would be apparent.
I think what people are asking is basically, could someone convert this to a program that is actually meant to be read, while still maintaining its exact control flow and logic.
for (B = i = y = u = b = i = 5 - 5, x = 10, I = [], l = []; l[B] = "ECDFBDCEAAAAAAAAIIIIIIIIMKLNJLKM@G@TSb~?A6J57IKJT576,+-48HLSUmgukgg OJNMLK IDHGFE".charCodeAt(B) - 64, B++ < 120; I[B - 1] = B % x ? B / x % x < 2 | B % x < 2 ? 7 : B / x & 4 ? 0 : l[i++] : 7);
function X(c, h, e, s) {
e ^= 8;
for (var o, S, C, A, R, T, G, n, N = -1e8, O = 20; ++O < 99;) if ((o = I[T = O]) && (G = o ^ e) < 7) {
A = G-- & 2 ? 8 : 4;
C = o - 9 ? l[61 + G] : 49;
do
if (!(R = I[T += l[C]]) && !! G | A < 3 || (R + 1 ^ e) > 9 && G | A > 2) {
if (!(R - 2 & 7)) return 78 - h << x;
n = G | (e ? T > 29 : T < 91) ? o : 6 ^ e;
S = (R && l[R & 7 | 32] - h - G) + (G ? 0 : n - o ? 55 : (A < 2) + 1);
if (c > h || 1 < c & c == h && S > 2) {
I[T] = n;
I[O] = 0;
S -= X(c, h + 1, e, S - N);
if (!(h || c - 1 | B - O | T - b | S < -1e4)) return W(B = 0), e && setTimeout("X(2,0,8),X(1,0,8)", 50);
I[O] = o;
I[T] = R
}
if (S > N || !h & S == N && Math.random() < .5) if (N = S, c > 1) if (h ? s - S < 0 : (B = O, b = T, 0)) return S
}
while (!R & G > 2 || (T = O, (G || A > 2 | (e ? O > 78 : O < 41) & !R) && ++C--A))
}
return N
}
function W() {
i = "<table>";
for (u = 18; u < 98; i += ++u % x - 9 ? "<th width=60 height=60 onclick='I[b=" + u + "]>8?W(B=b):X(1,0,0)'style='font-size:50px'bgcolor=#" + (u - B ? u.9 & 1 || 9 : "d") + "0f0e0>&#" + (I[u] ? 9808 + l[67 + I[u]] : 160) + ";" : u++ && "<tr>");
document.body.innerHTML = i + "</table>"
}
W();
I used to get repeatedly trounced by "Video Chess" on the TI-99/4A. Don't know how large that game was, but being from '79 it couldn't have been that large. Not that I was (or am) any good at chess -- if I fired up that game under MESS today it would probably still trounce me.
As someone who have been involved with chess engine community for the last 15 years or so (not anymore) I can tell you that, until recently, almost every single new engine, that has been written from scratch instead of copying an OS implementation, almost always left out en passant and castling in its first few releases. Even when they did implement those rules in to the engine, they did so grudgingly or as an after thought. I wonder why.
Now, everyone just copies the few dozens of OS options available out there and use their move generator which support all the rules, they just tweak with the piece values and algorithm to give a "personal touch" to the engine.
Nice, but it isn't very good, lost to Fools Mate (it took a knight then got mated) (1 look ahead would see it, queen and bishop just attack pawn near king)