A move generator that I wrote in Java has a special-purpose "test move" function that makes just enough of the move to allow bitboard attacks to be computed. It doesn't change the player, full/half move clocks, castling privileges, or Zobrist hash: all it does is update the piece bitmaps so that ...

The so-called "super-piece" method seems simplest and is what I use. generate bishop attacks from the king square and intersect this bitmap with the opponent bishops/queens. Non-zero = in check. Repeat for rook attacks and opponent rooks/queens. Then knights and pawns are straightforward. I don't ...

I'm using magic bitboards at the moment, but I am keeping a piece array for lookups based on square. It's at least theoretically possiblre to have more than 2 bishops/knights/rooks as well, but still only 15 pieces total, which is a definite improvement. I'll take a look into this.

I'm currently writing a chess engine, and have been trying to optimize the move generation before I got into the more complicated bits. I'm using bitboards with a piece lookup by square table. For check validation, I'm waiting until after the move has been made, then looking at the attacks on the ...