Bug: As white, if the a1 rook enters f1 or g1 during the game it forfeits short castling rights.

Example: http://en.lichess.org/AxeBpKcd2Fj3 ...here white should be able to castle on move 10.

I only tried this out as white and on short castling so the same bug might be present as black and in long castling.

what a bizar bug.

That's actually fascinating

I'm a retired software engineer. I see how that could happen. A flag is set on the square, indicating that a rook - any rook - has been on f1 or g1, rather than a flag set for the piece, ie the king's rook, indicating if it has moved. Using flags on the squares makes sense, since you want to check the squares for being under attack, or occupied by a piece of either color. The fix could be a bugger.

Nice find :o

never go fullretard TexasHoldem NEVER!

Nice analysis owltune.
This is a consequence of the application of murphy's law.
It happens whenever like here when the developper wants to make the code simpler, for different reasons, than the exact rule.

Pretty sure you mean a direct consequence of Morphy's Law.

More interesting is 9. Kf1!

Whoa, yeah, that's a heck of a move! :D