1. At least when analyzing a match, if the move played is the worst (the last one) the 'number of the move' (1, 2, 3, etc...) is shown as -1. 2. Both in 3D and 2D, if you click in another move that was previously played before gnubg had finished it's move (display computer move on) continues moving although the position changes. 3. Playing with the tutor against gnubg, if you rollout a move and then analyze the session, the rollout is erased. This don't happen with cube decisions nor if you analyze before rolling out and again afterwards. I thought this was a consequence of not having the same settings for analysis and evaluation, but choosing world-class in both showed the same behavior. 4. Also with the tutor, wrong cube decisions are often not marked (mainly drops and takes).
1. Couldn't reproduce playing against Gnubg, but I can send a game when that happens. 2. Not very easy to reproduce: (and even more difficult to explain) - Display computer moves must be on, and it's better if the speed isn't too fast. - Play some moves against Gnubg. - Edit a position where gnubg has the board closed, a couple of checkers in the outfield to move and owns the cube, and the other player has one or more checkers in the bar. - Let Gnubg play it's turn setting any dice. - Gnubg will play, the player will dance and the dice will be picked up. In the moment when the dice is picked up click in a *previous* move. To make it easier, double-click the move since the moment when the dice are being thrown. - The program will repeat what had been moved before the move selected. 3. Start a session, play anything, make any rollout of the move and analyze the session. The rollout results will disappear. This doesn't happen if you rollout a Gnubg's move. 4. Tutor must be on. Set this position: 4HPwASiw58EBAw:MAEAAAAAAAAA. Let Gnubg play it's turn (will double). Drop anyway, the action won't be marked. By the way, my OS is Windows Vista. Joaquin. PD: Is it possible to multi-thread the evaluation of just one move? For example, if you evaluate 8/5 6/5 in 4-ply only one core is used.
_______________________________________________ Bug-gnubg mailing list Bug-gnubg@gnu.org http://lists.gnu.org/mailman/listinfo/bug-gnubg
