Re: [gnugo-devel] a bug

[Paul Pogonyshev]

Wolfgang wrote:
> [skipped]
>
> ***assertion failure:
> board.c:964 - pos == PASS_MOVE || board[pos] == EMPTY near C5***
>
> it seems to me the reason is that in a sgf sequence:
>
> ;A;B(;C
>
> the parent of C comes out as A and it should be B.
>
> you are alright, since you never use SGFNode *parent
>
> (except in goto ...)
>
> i can write a fudge around that, but i guess one should really fix that

Well, it seems the following thingie should do the trick, though i admit
i haven't tested it, because i'm running regressions right now.

Paul



--- sgfnode.c.~1.24.~   2003-07-18 18:59:22.000000000 +0000
+++ sgfnode.c   2004-01-11 22:14:30.000000000 +0000
@@ -1049,7 +1049,7 @@ gametree(SGFNode **p, SGFNode *parent, i
     last = sequence(head);
     p = &last->child;
     while (lookahead == '(') {
-      gametree(p, last->parent, STRICT_SGF);
+      gametree(p, last, STRICT_SGF);
       p = &((*p)->next);
     }
     if (mode == STRICT_SGF)
@@ -1105,7 +1105,7 @@ gametreefuseki(SGFNode **p, SGFNode *par
        break;
       }
       else {
-       gametreefuseki(p, last->parent, mode, moves_per_game, i);
+       gametreefuseki(p, last, mode, moves_per_game, i);
        p = &((*p)->next);
       }
     }