Daniel Shawul wrote:Hello Rein
I have fixed it now and is uploaded. It was still a problem with declaration of variables . I am sure v1.0 had all of this working correctly since I had matching perft numbers. Anyway thanks for the report. I appreciate it.
Note that the 10x10 perft numbers have duplicate captures removed (i.e. when capturing 4 men you sometimes can do this clockwise and counterclockwise, this is then counted as 1 unique move).
Yes that was exactly what I used. I forgot that I saw your name somewhere My perft numbers match exactly for 8x8 checkers. For international checkers
it matches upto only depth=8. For depth 9 upwards, there are some circular captures (>=4) i.e different paths but same piceces captured, that I do not weed out in my move generator. Because of that it does not match but I think it is safe.
Since I use piece lists, the speed of move generation is much much slower than yours. This probably won't change soon since my goal is to try out different types of games where bitboards are not suitable.
Bug: Winboard 4.5 rejects move in Go and reversi, saying " Xboard: ...bare king" Winboard 4.4 works with the new version so it is a problem with the new version of winboard.
I uploaded a new version, which now implements 8 colors (+T):
RGBYCMWD
The colors are also no longer coupled to highlightColor and premoveHighlightColor, but R is a hard red, and Y a hard yellow.
The native -showTargetSquares option now uses G (green) for marking capture victims/to-squares that you can capture with other pieces than the one you lifted in suicide/losers. So if you plan to do something similar in Checkers, please stick to that. The other colors are not used in native highlight mode. Use of W (white) and D (black) is not recommended, as they look too much like the default piece colors.
The new version only reacts to highlight commands when native highlighting is off (i.e. -showTargetSquares false OR -testLegality false). It only reacts to click commands when one-click moving is on AND legality testing is off. (With the latter on it would use its native one-click moving.)
Daniel Shawul wrote:Bug: Winboard 4.5 rejects move in Go and reversi, saying " Xboard: ...bare king" Winboard 4.4 works with the new version so it is a problem with the new version of winboard.
What exacty does it say? Looking at the source it seems a message containing "bare king" should only ever be sent in Shatranj or Losers, or when you loseon time. I guess this is an adjudication? Just switch off all adjudications.
The coloring is changed to Red, Yellow and Green now. It works perfectly.
I am going to use one more color to instead of small r to differentiate b/n the different leg destination squares.
The first engine makes the first move, but the second's engine moves is rejected. And it is not coming from the first engine. I also turned off all adjuncations.
<000000000516> accepted option
<000000000516> accepted done
<000000000531> new
<000000000547> random
<000000000547> variant 9x9+0_go
setup (P.....p.....) 9x9+0 9/9/9/9/9/9/9/9/9 w - - 0 1
<000000000547> level 40 5 0
<000000000547> post
<000000000547> hard
<000000000547> easy
<000000000547> force
<000000000547> setboard 9/9/9/9/9/9/9/9/9 w - - 0 1
<000000000578> computer
<000000000578> name NebiyuGo_1.4
Hello NebiyuGo_1.4!
<000000007688> time 30000
<000000007688> otim 29285
<000000007688> usermove P@g4
<000000007688> go
[st = 7088ms, mt = 149250ms , moves_left 40]
9/9/9/9/9/6P2/9/9/9 b - - 0 1
[Tree not found]
Time : 7125ms Tree : nodes 65767 depth 9/86 pps 5361 visits 38200
Best move : g3 from 81 out of 81 moves [100.00%]
<000000014828> move p@g3
<000000014828> result 1-0 {Xboard: Forfeit due to invalid move: p@g3 (a1a1) res=24}
Well, this is quite a different error message than you first reported. 'Invalid move' means it could not be parsed. (Unlike 'illegal move', which means the move is understood by WB, but considered illegal, which obviously can only happen with legality-testing on.)
Couldn't it be the problem that you write a lower-case 'P'?
That 4.4 is different from the current version should not be a surprise, considering that the move parser has been totally rewritten from scratch. This is the main reason WB can handle double-digit rank numbers now.
Lower-case piece in drop moves is non-compliant. WB protocol prescribes upper case.
That 4.4 is different from the current version should not be a surprise, considering that the move parser has been totally rewritten from scratch. This is the main reason WB can handle double-digit rank numbers now.
Lower-case piece in drop moves is non-compliant. WB protocol prescribes upper case.
Oh I didn't know I can play 19x19 now. It works. Does the rank still starts with 0 ? Anyway I will test it myself.
Ok I will stick with the protocol and capitalize all letters before sending to winboard. I will have to change the parser to get the color of the pawn from the side to move not from the capitalization.
Edit:
There are some weird pieces in the middle of the 19x19 go board at startup
It seems there are some issues. The whole board got filled with white pieces on my first click.
My perft numbers match exactly for 8x8 checkers. For international checkers