gnugo-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [gnugo-devel] Re: bug?


From: bump
Subject: Re: [gnugo-devel] Re: bug?
Date: Sun, 23 Feb 2003 20:50:55 -0800

> I have one more question.  I wish to show the dead stones
> after the game is over.
> 
> I know i should be using gtp... But I'm not.
> 
> if you were to:
> 
> sh$ wget http://www.voltar.org/gnugo/sgf/105.sgf
> sh$ echo dead | gnugo --level 1 --quiet --mode ascii -l 105.sgf
> 
> It shows all stones as dead...  :(

The patch at the end should help.

Why are you running at level 1? At level 1 you cannot expect
any accuracy in the answer to this question.

One possible flaw that I see in your approach (I'm only
guessing at what you do, correct me if wrong) is that
if each move is generated by a separate gnugo process then
you are not taking advantage of the persistent caching.
This means that the game will progress a lot more slowly.

Anyway you can't expect much at level 1.

> If you were then to:
> 
> sh$ gnugo --level 1 --quiet --mode ascii -l 105.sgf
> 
> black(58): force pass
> white(59): force pass
> 
> You'd get:
> 
> Segmentation fault

This causes estimate_score to get run without first
examine_position and causes a crash. This patch doesn't
address that.

Dan

Index: play_ascii.c
===================================================================
RCS file: /cvsroot/gnugo/gnugo/interface/play_ascii.c,v
retrieving revision 1.36
diff -u -r1.36 play_ascii.c
--- play_ascii.c        22 Feb 2003 10:54:24 -0000      1.36
+++ play_ascii.c        24 Feb 2003 04:28:17 -0000
@@ -831,6 +831,7 @@
            current_score_estimate = NO_SCORE;
          break;
        case CMD_DEAD:
+         examine_position(gameinfo->to_move, FULL_EXAMINE_DRAGONS);
          showdead = !showdead;
          break;
        case CMD_CAPTURE:











reply via email to

[Prev in Thread] Current Thread [Next in Thread]