Booot 6.2 POP x64 problem ... epd can't load!

[Gusev]

Hi Frank,

From the same data, can you generate a CTG book, which would be useful when playing in the Engine mode on InfinityChess? For example, a drawish line that you normally exclude may still be sometimes better than an imbalanced line where an engine is more likely to lose. But if there is an alternative line giving more advantage, then the drawish one should be rejected.

Dmitri

[Werner]

Hi,
as Booot accepts other positions there could be a bug in the engine.
Why not send an email to the author?

Code: Select all

### 13.: to Booot 6.2_x64_popcnt (0): debug on

### from Booot 6.2_x64_popcnt (0): id name Booot 6.2_x64_popcnt

### from Booot 6.2_x64_popcnt (0): id author Alex Morozov (booot76@gmail.com)

### from Booot 6.2_x64_popcnt (0): option name Hash type spin default 128 min 16 max 8192

### from Booot 6.2_x64_popcnt (0): option name Ponder type check default false

### from Booot 6.2_x64_popcnt (0): option name Threads type spin default 1 min 1 max 16

### from Booot 6.2_x64_popcnt (0): uciok

setoption name Hash value 256

### 14.: to Booot 6.2_x64_popcnt (0): setoption name Hash value 256

setoption name Ponder value false

### 15.: to Booot 6.2_x64_popcnt (0): setoption name Ponder value false

### 16.: to Booot 6.2_x64_popcnt (0): isready

### from Booot 6.2_x64_popcnt (0): readyok

Engine: Booot 6.2_x64_popcnt (256 MB)
von Alex Morozov (booot76@gmail.com)
### 17.: to Booot 6.2_x64_popcnt (0): ucinewgame

### 18.: to Booot 6.2_x64_popcnt (0): isready

### from Booot 6.2_x64_popcnt (0): readyok

### 19.: to Booot 6.2_x64_popcnt (0): position fen rnbr2k1/pp3p2/1qp3pp/4Q1N1/4P3/2P3P1/PP3PBP/R3R1K1 w - -

### 20.: to Booot 6.2_x64_popcnt (0): go movestogo 40 wtime 14640000 btime 14640000

### from Booot 6.2_x64_popcnt (0): info depth 16 score cp 90 time 281 nodes 336672 nps 1198120 pv  g5f3 c8g4 a1d1 b8d7 e5f4 g4f3 g2f3 d7c5 f4e3 c5e6 e3b6 a7b6 d1d8 a8d8 e1d1 d8d1 f3d1 b6b5 f2f4

### from Booot 6.2_x64_popcnt (0): info depth 16 time 281 nodes 337111 nps 1199683

### from Booot 6.2_x64_popcnt (0): info depth 17

### from Booot 6.2_x64_popcnt (0): info depth 17 score cp 89 time 360 nodes 452471 nps 1256863 pv  g5f3 c8g4 e5f4 g4f3 g2f3 b8d7 e1e2 d8e8 e2d2 d7e5 f3e2 h6h5 a1d1 a7a6 a2a3 a8c8 g1h1

[hgm]

The UCI description at WBEC does not specify anything about the FEN.

But the FEN spec includes the halfmove and move counters, doesn't it? If you send something that is not compliant with the FEN spec, it is not really a FEN.

Strange as it may be, Shredder GUI is not compliant with the UCI specs.

as Booot accepts other positions there could be a bug in the engine.

Code: Select all

### 19.: to Booot 6.2_x64_popcnt (0): position fen rnbr2k1/pp3p2/1qp3pp/4Q1N1/4P3/2P3P1/PP3PBP/R3R1K1 w - -

The difference with the other case doesn't seem so much that it is another position, but that in this case there is no "moves" keyword behind the FEN. Apparently Booots FEN reader doesn't mind a FEN being truncated, but it does get confused when other info appears on the same line before the FEN is finished. As the log shows it then starts to emit (fully justified) error messages against the non-compliant stuff.

[Frank Quisinsky]

Hi Werner,

I have no interest to send mails to the authors.
Do that in Winboard times, Arena times or SWCR times more as 1.000x with bug reports and so on. This time ended for myself ... because no interest in beta testing. If the programmer not found the message here in TalkChess ... it's fully OK for me.

If an engine make problems ... very easy ... the next on.

After all the years computer chess to tired for such things. Honestly should the others do. I do here enough. Furthermore, I am to old for such things, I make to many mistakes comparing for 10 years. Haven't the energy today.

If I am ready with FEOBOS (my last project in computer chess) I have four engines on my system for self playing / analysis / engines matches. This is enough. And believe me ... with this 4 programs I can do all what I like and have many fun. One of them is Wasp I like a lot.

Booot is really a very interesting engine but I have no interest to lost to many time for FEOBOS because the project runs long enough.

Best
Frank

[Frank Quisinsky]

Hi Dmitri,

thats right.
But all project files are open ...
EPD, PGN to each position a position number and with Excel workbook users can set to all important facilities for alteration an own setting.

With other words:
Loading the PGN in Chessbase and select out what you dislike, the Excel file give you.

The files I public on my site are with my own favorite settings. Must not be the best.

That's the deal ... all should be open and should give the others possibilities for own changes.

I hope that with FEOBOS files later the others added own ideas and make the main idea better I do. The Excel workbook helps here.

Best
Frank

To select out the 0,00 draws is a bit critical. Not all of that lines ended with a fast 3-fold repetition. But I am thinking that to the end of project I reject if 40% of engines gave a 0,00 evaluation. With the deal that the book have a very high contempt for fast draws. 70% of such fast draws can be avoided but many other 0,00 draw variants, not ended in 3-fold repetition will be lost.

So, I am searching since a while for more balanced lines, 3 moves after ECO code was formed and found so far over 22.000 never checked new variants. At the moment not checked for mistakes ... maybe 14.000 good new variants.

If I have later around 35.000 balanced lines to all ECO codes I am happy.

Most of the new lines are theory relevant from 1940 - 1990 and games from the last 10 months. Can be found on my FEOBOS sites since yesterday. After engines checked the current 26.146 lines I am starting directly an update.

For that reason I will delete 0,00 draw lines stricter because end of the project I have enough in the book.

[Guenther]

The UCI description at WBEC does not specify anything about the FEN.

But the FEN spec includes the halfmove and move counters, doesn't it? If you send something that is not compliant with the FEN spec, it is not really a FEN.

Strange as it may be, Shredder GUI is not compliant with the UCI specs.

as Booot accepts other positions there could be a bug in the engine.

Code: Select all

### 19.: to Booot 6.2_x64_popcnt (0): position fen rnbr2k1/pp3p2/1qp3pp/4Q1N1/4P3/2P3P1/PP3PBP/R3R1K1 w - -

The difference with the other case doesn't seem so much that it is another position, but that in this case there is no "moves" keyword behind the FEN. Apparently Booots FEN reader doesn't mind a FEN being truncated, but it does get confused when other info appears on the same line before the FEN is finished. As the log shows it then starts to emit (fully justified) error messages against the non-compliant stuff.

My theory is that S12 adds that strange 'moves' string (for whatever reason), when all castling rights are delivered.
This would explain why it happened not very often in Franks epd as it holds mostly longer but variable opening
lines. I haven't done a test for verification of my idea.

I sent a PM to Alex (Booot), but it seems he doesn't read here very often.
No idea if SMK still reads here ;-)

Guenther

[Frank Quisinsky]

Hi Günther,

no explanation from my site:
But here are two other examples:

r1bqkb1r/p1p3p1/2P1pn1p/1p2p3/8/8/PPPP1PPP/RNBQKB1R b KQkq - id "14377";

r2qk2r/1ppn1ppp/p1pbb3/4p3/4P3/1P1P1N2/P1PN1PPP/R1BQ1RK1 w kq - id "15518";

I stopped with position 15517!

Best
Frank

[hgm]

Well, position ... moves ... is standard UCI syntax. Many interfaces send 'moves' even if the move list is empty. UCI2WB does this, and I think Polyglot does it too. I never heard any complaints about it. The UCI specs do not suggest that 'moves' should only be sent when the move list is non-empty.

It is a bit strange that Shredder GUI doesn't do it consistently.

Is Shredder GUI to blame for the clipping of the halmoves and move-number fields, or was the input start position already an EPD without these fields? Perhaps the problem is that Shredder GUI requires FENs, and feeding it EPDs is wrong, leading to undefined results.

[booot]

Booot does not understand what does string
"position fen rnb1k2r/pppnqppp/4p3/3pP3/3P1P2/2N5/PPP3PP/R2QKBNR b KQkq - moves" mean. Strange word "moves" here (while he waits move counter like descripted in standart) without any information about moves.
I did FEN parser code many years ago, so , may be the time to revisit it now and make it more protected.

Alex.

[hgm]

Obviously Booot is not to blame. It is either a Shredder bug or user error.