gnugo-devel
[Top][All Lists]
Advanced

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

Re: [gnugo-devel] Another bug report


From: Gunnar Farneback
Subject: Re: [gnugo-devel] Another bug report
Date: Tue, 01 Jan 2002 18:39:27 +0100
User-agent: EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/20.7 (sparc-sun-solaris2.7) (with unibyte mode)

(I move this reply to the list.)

Markus wrote:
> Here is another failure of GnuGo, this time reproducible.

Thanks.

> gnugo -r 1 --mode gtp --chinese-rules --capture-all-dead --cache-size 48
> 
> In one of the games GnuGo tried to play a move violating the Ko rule.
> The game is appended to this mail, after the last move, GnuGo tried
> to play J2 as Black.

Confirmed. This particular problem can only occur when the
--capture-all-dead option is used, which we haven't tested very much.
The appended patch for GNU Go 3.0.0 fixes it. GNU Go 3.1.18 doesn't
have this problem but that turned out to be due to a (different) bug.
:-)

/Gunnar

--- aftermath.c.orig    Tue Jan  1 18:15:19 2002
+++ aftermath.c Tue Jan  1 18:16:32 2002
@@ -709,7 +709,9 @@
              || do_capture_dead_stones)
          && (DRAGON2(m, n).safety == DEAD
              || DRAGON2(m, n).safety == TACTICALLY_DEAD)
-         && worm[m][n].attack_code != 0) {
+         && worm[m][n].attack_code != 0
+         && !is_illegal_ko_capture(worm[m][n].attacki, worm[m][n].attackj,
+                                   color)) {
        *i = worm[m][n].attacki;
        *j = worm[m][n].attackj;
        return 1;



reply via email to

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