Major bug in ChessGUI 0.185, 0.186, 0.187

[Matthias Gemuh]

There is a major bug in ChessGUI versions 0.185, 0.186, 0.187. !

When a programmer mistakenly types

Code: Select all

if ((tc->nChoice<0)||(tc->nChoice>0)) tc->nChoice = 0;

instead of

Code: Select all

if ((tc->nChoice<0)||(tc->nChoice>2)) tc->nChoice = 0;

, the consequences can be very devastating.

Time control is then wrongly always fischer.

I will upload ChessGUI 0.188 later today.


Matthias.

[Matthias Gemuh]

ChessGUI 0.188 is out there.

- fixed repeated TC bug
- Tournaments with more than 32 cycles (usually gauntlets) will now initialize faster (by postponing some initialisations)
- Options for creating GUI Remis Book now also used to create pgn start positions test suites.


Matthias.

[pedrox]

ChessGUI 0.188 is out there.

- fixed repeated TC bug
- Tournaments with more than 32 cycles (usually gauntlets) will now initialize faster (by postponing some initialisations)
- Options for creating GUI Remis Book now also used to create pgn start positions test suites.


Matthias.

Now, I have problems with chess960, with version 0.181 no.

[Matthias Gemuh]

Matthias.
Now, I have problems with chess960, with version 0.181 no.

Maybe it is a DanaSah problem ?

I have just let BrightFRC-0.4a play a Chess960 gauntlet against

Code: Select all

Daydreamer 1.75 JA
Danasah 4.45      
Jonny 2.83        
Baron 1.7 FRC     
DanaSah 4.20      
Hermann 2.5       

under ChessGUI 0.189 using positions

Code: Select all

bqnbnrkr/pppppppp/8/8/8/8/PPPPPPPP/BQNBNRKR w KQkq - 0 1
bqnnrbkr/pppppppp/8/8/8/8/PPPPPPPP/BQNNRBKR w KQkq - 0 1

The 2 DanaSahs always play the illegal move a8a8 as first move when playing as black.
As white, they have no problems.

All the other engines in this tournament have no problems.

TimeControl : Game in 1 minute.

The 2 DanaSahs are configured in their ini files to use chessgui and castle with O-O .

Matthias.

[pedrox]

Matthias.
Now, I have problems with chess960, with version 0.181 no.

Maybe it is a DanaSah problem ?

I have just let BrightFRC-0.4a play a Chess960 gauntlet against

Code: Select all

Daydreamer 1.75 JA
Danasah 4.45      
Jonny 2.83        
Baron 1.7 FRC     
DanaSah 4.20      
Hermann 2.5       

under ChessGUI 0.189 using positions

Code: Select all

bqnbnrkr/pppppppp/8/8/8/8/PPPPPPPP/BQNBNRKR w KQkq - 0 1
bqnnrbkr/pppppppp/8/8/8/8/PPPPPPPP/BQNNRBKR w KQkq - 0 1

The 2 DanaSahs always play the illegal move a8a8 as first move when playing as black.
As white, they have no problems.

All the other engines in this tournament have no problems.

TimeControl : Game in 1 minute.

The 2 DanaSahs are configured in their ini files to use chessgui and castle with O-O .

Matthias.

It's weird. DanaSah 4.20 has played thousands of games without problems of chess960 in the CCRL list. I've had no problem when I use version 0.181 or lower. To play, I choose gaunlet and random per game, I have not tried starting positions. The program then plays well the white and black.

Now I've replaced version 0.181 by 0.188 version and play in the same conditions. The first engine start and move, where is the turn of the second engine displays the first line of thought that says "searching" and then there is a beep and the game is finish. If I return to the previous version everything is correct.

Perhaps among the versions has changed something on sending FEN position?

When I have more time I will study the problem.

Pedro

[pedrox]

In chess960, random per cycle always play the same positions all games.

[pedrox]

In chess960, random per cycle always play the same positions all games.

the above is for version 0181.

In the 0188 version by choosing random per cycle, all items have a different position, not cycle.

[pedrox]

The difference I see between version 0181 and version 0188:

First move for black.

version 0.181:
SendToEng2Time 0000000240253500 : Eng2 (danasah424_WB) -> time 6000
SendToEng2Time 0000000240253500 : Eng2 (danasah424_WB) -> otim 6000
SendToEng2Time 0000000240253500 : Eng2 (danasah424_WB) -> a2a4
SendToEng2Time 0000000240253500 : Eng2 (danasah424_WB) -> go
SideToMove = Black.

version 0.188:
SendToEng2Time 0000000290904984 : Eng2 (danasah453) -> f2f4
SendToEng2Time 0000000290904984 : Eng2 (danasah453) -> go
SideToMove = Black.

Now DanaSah not receive the command time and otime.

Best regards,

Pedro

[Matthias Gemuh]

In chess960, random per cycle always play the same positions all games.

the above is for version 0181.

In the 0188 version by choosing random per cycle, all items have a different position, not cycle.

I already noticed that in 0.188.
0.185 was a major update to simplify program logic and 0.189 will fix all known bugs.

[Matthias Gemuh]

The difference I see between version 0181 and version 0188:

First move for black.

version 0.181:
SendToEng2Time 0000000240253500 : Eng2 (danasah424_WB) -> time 6000
SendToEng2Time 0000000240253500 : Eng2 (danasah424_WB) -> otim 6000
SendToEng2Time 0000000240253500 : Eng2 (danasah424_WB) -> a2a4
SendToEng2Time 0000000240253500 : Eng2 (danasah424_WB) -> go
SideToMove = Black.

version 0.188:
SendToEng2Time 0000000290904984 : Eng2 (danasah453) -> f2f4
SendToEng2Time 0000000290904984 : Eng2 (danasah453) -> go
SideToMove = Black.

Now DanaSah not receive the command time and otime.

Best regards,

Pedro

Yes, ChessGUI 0.188 breaks protocol there. Will fix that in 0.189.

Thanks,
Matthias.