Handling of pseudo-final positions in bughouse

[antoyo]

Hi. It seems wrong to me that a stalemate is considered to be a checkmate in bughouse. The following commands:

setoption name UCI_Variant value bughouse
position fen 7k/P1Q5/8/8/3KP3/8/1PPP1PPP/RNB2BNR w - - 1 25
setoption name MultiPV value 15
go mate 1

gives the following mates in one:

info depth 245 seldepth 2 multipv 1 score mate 1 nodes 207151 nps 912559 tbhits 0 time 227 pv a7a8q
info depth 245 seldepth 2 multipv 2 score mate 1 nodes 207151 nps 912559 tbhits 0 time 227 pv a7a8r
info depth 245 seldepth 2 multipv 3 score mate 1 nodes 207151 nps 912559 tbhits 0 time 227 pv f1c4
info depth 245 seldepth 2 multipv 4 score mate 1 nodes 207151 nps 912559 tbhits 0 time 227 pv c7f7

In my opinion, the last two should not be considered as checkmate because they are stalemate. Thanks to fix this issue.

[ianfab]

Thanks for the report. This behavior was implemented intentionally, since OTB what happens in case of stalemate is that the player can not move and his time runs out. However, I do not know how stalemate is handled in online/computer bughouse, so input on that is very welcome.

Something similar actually applies to checkmates that still can be parried by a piece drop, but where the player does not have a piece in hand. On FICS I think such a situation immediately ends the game, whereas OTB the player might still wait for a piece from the other board if the opponent does not sit, so I am not sure whether that behavior is intentional or a limitation, and how it should be treated in Fairy-Stockfish.

[antoyo]

This website states that stalemates and non-contact checkmate do not end the game:

The player is not considered to be stalemated (since it is possible to get a piece to drop) The player is not considered to be checkmated if he could block the check with a piece potentially in his stock.

Also, I don't know for FICS, but on chess.com, a checkmate that can be blocked does not end the game immediately.

[ianfab]

Thanks. I think from an evaluation perspective it is ok to treat stalemate as winning, because it forces the opponent to sit until he gets pieces. A proper solution of both the situations with stalemate as well as with checkmates that can still be blocked depends on #122 to be implemented, but I will keep this open as a reminder to tackle this as soon as that issue is solved.