Mike Roberts wrote: > There seems to be a bug in GNU Chess 5.0 which caused it to lose the > attached game (with 23... Kh8) rather than settle for a draw by > perpetual check with (23...Kf8 24. Qe7+ Kg8 25. Qe6+ Kf8 and so on). It > seems strange that the program chooses to give white mate in two rather > than settle for a draw.
Thanks Mike, seems to be a problem even when the relatively recent "ponder" feature is disabled. It does consider Kf8 best to a depth of 32768 ply, before playing Kh8, and does seem to play Kf8 in some cases (first occurrence). So some sort of search termination problem. I know there are some edge cases when mate is imminent for one side or other, where the code goes wrong, but this looks like a useful test case. Simon
signature.asc
Description: OpenPGP digital signature
_______________________________________________ Bug-gnu-chess mailing list [email protected] http://lists.gnu.org/mailman/listinfo/bug-gnu-chess
