Re: [gnugo-devel] A problem in atari_atari

[Trevor Morris]

>The stone at B1 should not be classified as critical
>and I'll have to fix this in the semeai code. However
>it still seems to me that there will be a potential
>for a crash at line 641 of combination.c in error
>situations. Probably str should be tested for
>emptiness and if this happens the conservative conclusion
>would be to conclude that there is no combination.
>
>Thus I propose the following. This includes the previous
>patch.

Shouldn't there be a warning message issued to stderr, if not
an assert, as this should never happen?


>Dan
>
>
>Index: engine/combination.c
>===================================================================
>RCS file: /cvsroot/gnugo/gnugo/engine/combination.c,v
>retrieving revision 1.10
>diff -u -r1.10 combination.c
>--- engine/combination.c       5 Dec 2001 19:49:20 -0000       1.10
>+++ engine/combination.c       12 Dec 2001 18:24:31 -0000
>@@ -637,6 +637,12 @@
>     if (!trymove(apos, color, "do_atari_atari-A", str, EMPTY, NO_MOVE))
>       continue;
>       
>+    if (!IS_STONE(str)) {
>+      /* Error situation */
>+      popgo();
>+      return 0;
>+    }
>+
>     /* Try to defend the stone (str) which is threatened. */
>     aa_val = countstones(str);
> 
>Index: engine/reading.c
>===================================================================
>RCS file: /cvsroot/gnugo/gnugo/engine/reading.c,v
>retrieving revision 1.38
>diff -u -r1.38 reading.c
>--- engine/reading.c   10 Dec 2001 17:22:40 -0000      1.38
>+++ engine/reading.c   12 Dec 2001 18:25:21 -0000
>@@ -895,12 +895,10 @@
> 
>       for (r = -1; r < max_points; r++) {
>         /* -1 is used only when stackp > 0. */
>-        if (stackp > 0) {
>-          if (r == -1)
>-            bb = dd;
>-          else
>-            break;
>-        }
>+        if (r == -1 && stackp > 0)
>+          break;
>+        if (r == -1)
>+          bb = dd;
>         else {
>           if (worm[adjs[k]].defend_codes[r] == 0)
>             break;
>
>
>
>_______________________________________________
>gnugo-devel mailing list
>address@hidden
>http://mail.gnu.org/mailman/listinfo/gnugo-devel
>