Cute Chess 0.9.4 released

[hgm]

Sure. Just tick "use each line/position twice". That also works when both engines use GUI book.

[ilari]

Suggestion:
Implement zh epd file as opening suite.

EPD format:

Code: Select all

r1bqkbnr/pp3ppp/2ppp3/8/2BQP3/2N5/PPP2PPP/R1B2RK1[PNn] b kq -

There was a stupid bug that broke Crazyhouse FEN parsing. I just fixed it in the master branch. So now there should be nothing preventing the use of Crazyhouse EPD opening suites.

I still have a couple of bugs and minor improvements left before I release the next version. BTW, have you experienced any more crashes with Cute Chess 0.9.4?

[Ferdy]

Sure. Just tick "use each line/position twice". That also works when both engines use GUI book.

Not exactly what I wanted as it re-entered the book again.

[Event "Computer Chess Game"]
[Site "Computer II"]
[Date "2016.09.26"]
[Round "3"]
[White "Sunsetter 7g 64bit"]
[Black "Imortal 2.0"]
[Result "0-1"]
[TimeControl "60+1"]
[Variant "crazyhouse"]
[Annotator "6. +0.08 8... +2.51"]

1. e4 e5 2. Nf3 Nc6 3. Bc4 Bc5 4. O-O d6 5. d3 Nf6 6. Bg5 {+0.08/11 0.9}
Bg4 7. Nbd2 {+0.14/11 0.8} O-O 8. Re1 {+0.00/11 0.9} Bxf2+ {+2.51/9 3} 9.
Kxf2 {-0.05/11 0.9} Bxf3 {+2.49/9 3} 10. Nxf3 {-0.05/9 0.8} Ng4+
{+4.02/9 3} 11. Kf1 {-0.60/10 0.8} Nxh2+ {+4.35/9 2.8} 12. Nxh2
{-0.81/10 0.8} Qxg5 {+6.41/8 2.6} 13. B@e3 {-0.81/9 0.9} N@g3+
{+4.15/8 2.2} 14. Kg1 {-0.95/10 0.8} P@f4 {+5.55/9 2.2} 15. Bf2
{-1.40/10 1.7} B@d4 {+4.40/8 2.6} 16. Bxd4 {-2.15/10 0.8} exd4
{+6.79/9 2.6} 17. N@f2 {-1.13/10 3} P@e3 {+7.88/9 2.5} 18. Nhg4
{-4.09/11 13} exf2+ {+12.50/8 2.5} 19. Nxf2 {-4.62/11 0.7} N@e3
{+14.01/9 2.4} 20. B@f3 {-4.42/9 1.3} Nxd1 {+19.70/9 2.1} 21. Rexd1
{-14.93/11 10} B@e3 {+26.16/8 2.3} 22. N@e7+ {-14.93/9 0.6} Qxe7
{+299.89/9 2.1} 23. Rf1 Q@h1# {+299.99/5 0.4}
{Xboard adjudication: Checkmate} 0-1

[Event "Computer Chess Game"]
[Site "Computer II"]
[Date "2016.09.26"]
[Round "4"]
[White "Imortal 2.0"]
[Black "Sunsetter 7g 64bit"]
[Result "0-1"]
[TimeControl "60+1"]
[Variant "crazyhouse"]
[Annotator "6. +0.06 6... +0.19"]

1. e4 e5 2. Nf3 Nc6 3. Bc4 Bc5 4. O-O d6 5. d3 Nf6 6. Nc3 {+0.06/10 3} Bg4
{+0.19/12 1.6} 7. Bxf7+ {+1.17/10 3} Kxf7 {+1.09/10 0.8} 8. P@g5
{-0.92/9 3} Nh5 {+1.24/11 0.8} 9. h3 {+0.78/9 2.7} Bd7 {+0.87/10 1.6} 10.
Nxe5+ {+0.84/8 2.7} Nxe5 {+1.45/10 0.8} 11. Qxh5+ {+0.03/9 2.6} B@g6
{+1.35/11 1.6} 12. Qd1 {+1.24/9 2.6} N@h4 {+0.21/12 13} 13. N@a5
{+1.95/8 2.5} Bxh3 {+2.05/11 2.6} 14. P@g3 {-2.86/8 2.5} Bxg2
{+2.73/10 0.6} 15. gxh4 {-5.31/7 2.4} Bxf1 {+4.69/9 1.3} 16. Kxf1
{-8.02/7 2.4} R@h1+ {+5.33/9 0.7} 17. N@g1 {-11.46/8 2.3} P@h2
{+7.45/9 0.7} 18. Ne2 {-13.21/7 2.3} Nf3 {+13.49/9 0.7} 19. B@d5+
{-22.46/6 2.3} Ke7 {+299.20/10 0.7} 20. Kg2 {-299.88/8 2.2} Rxg1+
{+299.30/33} 21. Nxg1 {-299.90/11 1.8} N@e1+ {+299.40/33} 22. Qxe1
{-299.92/15 2.2} Nxe1+ {+299.50/33} 23. Kf1 {-299.94/14 2.1} hxg1=Q+
{+299.60/33} 24. Ke2 {-299.96/13 2.1} Qxf2+ {+299.70/33} 25. Kd1
{-299.98/4} Qxc2+ {+299.80/33} 26. Kxe1 {-299.98/3} Q@d1# {+299.90/33}
{Xboard adjudication: Checkmate} 0-1

[Ferdy]

Suggestion:
Implement zh epd file as opening suite.

EPD format:

Code: Select all

r1bqkbnr/pp3ppp/2ppp3/8/2BQP3/2N5/PPP2PPP/R1B2RK1[PNn] b kq -

There was a stupid bug that broke Crazyhouse FEN parsing. I just fixed it in the master branch. So now there should be nothing preventing the use of Crazyhouse EPD opening suites.

I still have a couple of bugs and minor improvements left before I release the next version. BTW, have you experienced any more crashes with Cute Chess 0.9.4?

I have not encountered any crashes anymore.

[hgm]

OK, I see. It cannot do that. It would only be possible anyway when you disable book after the first miss, because the games could have occurred in reverse order, so that you would transpose into the book in the second game. And if you would disable the book after the first miss, the situation can no longer occur.

I also don't see what you hope to achieve by this. Even if the first game would have engine A out of book much earlier than engine B, it would have been disadvantaged in that game by having to think about more moves, while in the second game its opponent would now get these moves for free.

[Ferdy]

OK, I see. It cannot do that. It would only be possible anyway when you disable book after the first miss, because the games could have occurred in reverse order, so that you would transpose into the book in the second game. And if you would disable the book after the first miss, the situation can no longer occur.

I also don't see what you hope to achieve by this. Even if the first game would have engine A out of book much earlier than engine B, it would have been disadvantaged in that game by having to think about more moves, while in the second game its opponent would now get these moves for free.

The use of book should be fair, re-entering is not allowed especially if the selection is random. Perhaps re-entering is allowed if the top move is always selected.

[Ferdy]

There was still a really nasty bug causing crashes in tournaments so I've made a bugfix release for 64-bit Windows and Linux: https://github.com/cutechess/cutechess/ ... /gui-0.9.4

There's also a bonus feature - support for the "UCI_Variant" UCI option.

Suggestion:
Provide a viewer for generated polyglot book, showing moves, weight, wins, loses.

[hgm]

The use of book should be fair, re-entering is not allowed especially if the selection is random. Perhaps re-entering is allowed if the top move is always selected.

Well, what you proposed certainly would not be fair, because (if I understood you correctly) the GUI would be feeding moves that one engine had to spend time on to think it up would be fed to the opponent as book move for free.

I still think the problem is with the book, in particular that you incorporate one-sided lines in it.

[Ferdy]

The use of book should be fair, re-entering is not allowed especially if the selection is random. Perhaps re-entering is allowed if the top move is always selected.

Well, what you proposed certainly would not be fair, because (if I understood you correctly) the GUI would be feeding moves that one engine had to spend time on to think it up would be fed to the opponent as book move for free.

I still think the problem is with the book, in particular that you incorporate one-sided lines in it.

The book was created from the games.
The suggestion was not to re-enter the book once it is out of book the first time, is this impossile to do?

[hgm]

Hardly anything is impossible. The question is whether it is a good thing having it. I could for instance make it such that when you set -bookDepth to -N (N>0) would be equivalent to setting it to +N, but would set a flag that disables book probing after the first miss. (And the flag would be cleared whenever a game starts.)