Author: F. Huber
Date: 03:41:59 01/23/05
Hello Odd-Gunnar, while playing around with the following position (posted yesterday) with TheKing installed as UCI-engine with your current WB2UCI-adaptor, I found a serious problem/bug in the last versions of your adaptor! [D]2q1k3/1p3p2/4P1pp/P1Qp3P/3K1P2/1P6/8/8 b - - If _you_ are make the first black move (e.g. Qxc5) in normal play mode, then TheKing doesn´t answer at all, it completely stops thinking after ply 3! Martin Baumung found out, that this happens at _all_ positions, where _black_ has to move first, if the user makes this first move and TheKing should make the next white move, and that the problem does _not_ occur with the quite old WB2UCI-adaptor version from 20. April 2002! So I´ve tried all versions from you download page and I can confirm, that this old 2002 version was the last one, which did not have this bug. After making logfiles from both versions (the old and the current one), I can see an important difference in the communication with the King-engine - have a look at the following part of these files: Old version: ----------- 8.371: C> position fen 2q1k3/1p3p2/4P1pp/P1Qp3P/3K1P2/1P6/8/8 b - - 0 1 moves c8c5 8.371: C> go wtime 900000 btime 885557 winc 0 binc 0 8.374: C< info depth 1 score cp 1 8.374: S< new 8.374: S< post 8.374: S< hard 8.374: S< force 8.374: S< setboard 2q1k3/1p3p2/4P1pp/P1Qp3P/3K1P2/1P6/8/8 b - - 0 1 8.374: S< level 0 15 0 8.374: S< c8c5 8.374: S< time 90000 8.374: S< otim 88555 8.374: S< white 8.374: S< go 8.388: S> 3002 +87 1 348 Kxc5 gxh5 exf7+ Kxf7 Kxd5 8.404: C< info depth 3 score cp 87 time 10 nodes 348 nps 34800 pv d4c5 g6h5 e6f7 e8f7 c5d5 8.471: S> 4003 +87 9 3813 Kxc5 gxh5 exf7+ Kxf7 Kxd5 8.489: C< info depth 4 score cp 87 time 90 nodes 3813 nps 42366 pv d4c5 g6h5 e6f7 e8f7 c5d5 8.509: S< . 8.530: S> stat01: 15 6144 4003 1 3 Kd3 Qc3+ Kxc3 b6 exf7+ Kf8 hxg6 d4+ Kxd4 ..... New version: ----------- 9.087: C> position fen 2q1k3/1p3p2/4P1pp/P1Qp3P/3K1P2/1P6/8/8 b - - 0 1 moves c8c5 9.087: C> go wtime 900000 btime 858269 winc 0 binc 0 9.087: C< info depth 1 score cp 1 9.087: S< force 9.087: S< new 9.087: S< post 9.087: S< hard 9.087: S< setboard 2q1k3/1p3p2/4P1pp/P1Qp3P/3K1P2/1P6/8/8 b - - 0 1 9.087: S< level 0 15 0 9.087: S< time 90000 9.087: S< otim 85826 9.087: S< c8c5 9.137: S> 3001 -7 4 1009 Qxe6 Qb5+ Qd7 Qxd7+ Kxd7 hxg6 fxg6 Kxd5 9.137: C< info depth 3 score cp -7 time 40 nodes 1009 nps 25225 10.090: S< . 11.091: S< . 12.095: S< . ..... I think the problem is in the following 2 commands "white" and "go", which are sent to the engine in the old version, but _not_ in the newer ones!? I hope you´re able to reproduce this bug and find the reason for it, and of course I´m looking forward to a new corrected WB2UCI version soon ... ;-) Best regards, Franz.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.