Not optimal Position.compute_winning_position

[yurkovas]

I'm removed (commented) the possibly unnecessary code in Position.compute_winning_position:

//horizontal
position_t p = (position << (HEIGHT + 1)) & (position << 2 * (HEIGHT + 1));
r |= p & (position << 3 * (HEIGHT + 1));
// r |= p & (position >> (HEIGHT + 1));
p = (position >> (HEIGHT + 1)) & (position >> 2 * (HEIGHT + 1));
// r |= p & (position << (HEIGHT + 1));
r |= p & (position >> 3 * (HEIGHT + 1));

//diagonal 1
p = (position << HEIGHT) & (position << 2 * HEIGHT);
r |= p & (position << 3 * HEIGHT);
// r |= p & (position >> HEIGHT);
p = (position >> HEIGHT) & (position >> 2 * HEIGHT);
// r |= p & (position << HEIGHT);
r |= p & (position >> 3 * HEIGHT);

//diagonal 2
p = (position << (HEIGHT + 2)) & (position << 2 * (HEIGHT + 2));
r |= p & (position << 3 * (HEIGHT + 2));
// r |= p & (position >> (HEIGHT + 2));
p = (position >> (HEIGHT + 2)) & (position >> 2 * (HEIGHT + 2));
// r |= p & (position << (HEIGHT + 2));
r |= p & (position >> 3 * (HEIGHT + 2));

[yurkovas]

I was wrong. compute_winning_position works correctly.