Game continues after being allowed to capture king

[harroun]

Playing as black against white computer (engine strength at 4 stars)

1. Nc3 e5
2. Nf3 Nc6
3. d3 Nf6
4. Be3 d5
5. Qd2 d4
6. Nd4 d4
7. Bd4 Nd4
8. e3 Qe7
9. d4 Qe1

[alaskalinuxuser]

Thanks for playing JustChess! I will have a look at this tomorrow. I will have to set up a board to look at the problem. It appears there is an error that did not stop the game at check mate. Can I ask, which version do you have installed? Thanks for bringing this up!

[harroun]

Version 1.7

[alaskalinuxuser]

At first, I thought there was a problem with the pawn function. I thought that it was not properly moving the pawn so it could check if the king was safe. Or perhaps that it was checking if the king was safe BEFORE moving the pawn. However, I realized that the pawn code didn't have any issues.

Then I thought that the king safe method was improper for determining if the king was actually safe. Yet, it was acurately determining that the king was safe, per the code I was looking at. So what could it be? Well, to solve that, I started doing some debugging by pulling logs. Particularly, I started pulling logs from each check of pawn moves. Here was the code I used:

Log.i("WJH pawn list", list + " " + whiteTurn + " " + whiteKing + " " + blackKing );

Which would return the pawn list of available moves, whose turn it was, and where the white and black kings were, respectively.

This was what I got in return:

05-31 12:24:41.591 19249-19249/com.alaskalinuxuser.justchess I/WJH pawn list: P1430*,P1422*, true 4 60
05-31 12:24:41.591 19249-19249/com.alaskalinuxuser.justchess I/WJH pawn list: P1531*,P1523*, true 4 60
05-31 12:24:41.591 19249-19249/com.alaskalinuxuser.justchess I/WJH pawn list: P1927*, true 4 60
05-31 12:24:45.126 19249-19249/com.alaskalinuxuser.justchess I/WJH pawn list: P0824*,P0816*, true 2 60
05-31 12:24:45.126 19249-19249/com.alaskalinuxuser.justchess I/WJH pawn list: P0925*,P0917*, true 2 60
05-31 12:24:45.126 19249-19249/com.alaskalinuxuser.justchess I/WJH pawn list: true 2 60
05-31 12:24:45.126 19249-19249/com.alaskalinuxuser.justchess I/WJH pawn list: true 2 60
05-31 12:24:45.126 19249-19249/com.alaskalinuxuser.justchess I/WJH pawn list: P1430*,P1422*, true 2 60
05-31 12:24:45.126 19249-19249/com.alaskalinuxuser.justchess I/WJH pawn list: P1531*,P1523*, true 2 60
05-31 12:24:45.126 19249-19249/com.alaskalinuxuser.justchess I/WJH pawn list: true 2 60

Whoa! Notice that the white king is on square #4 (E1) the whole game, but when it comes time for this move, the white king has "transported" himself to square #2 (C1)! So yes, the king is safe, because he's not there! Notice from this screenshot, after making the move in question, the black queen kills what looks like the white king. Then, as whites turn, the only valid moves are to defend the white king by either a. killing the black queen, or b. blocking the black queens threat on the white king who is "invisible" on square #2! Crazy! so the real question is not how can you kill the enemy king, nor how can you continue to play if the king is dead, but rather, how did the king become invisible and transport to square C1?

[Keisial]

It probably thought it castled queenside. Maybe it evaluated the position after castling, then "moved back" only the rook? However, it then happily captures Rxe1 despite the invisible king.

FWIW, from a similar position, it also moves away the bishop leaving the king in check, so it is clearly not pawn-specific.

[alaskalinuxuser]

Great observations! Yes, it is not pawn specific. There are some unusual things that happen with castling, so that may be where to look.

[alaskalinuxuser]

I can also see that this only appears to happen if the potential to castle exists, leading back to the code for castling......