Re: [gnugo-devel] Bug report gnugo-3.3.8

[Evan Berggren Daniel]

On Thu, 12 Sep 2002, Arend Bayer wrote:

> On Tue, 10 Sep 2002, Evan Berggren Daniel wrote:
>
> > > Markus wrote:
> > > > ***assertion failure:
> > > > board.c:909 - board[pos] == EMPTY near H8***
> >
> > Isn't this similar to the illegal plays I was reporting with
> > matcher_check?  Those are reproducible, I sent an earlier email that
> > includes a random seed that causes the illegal result.
>
> I see no similarity. I had added a link to your message from devel.html,
> but now I think it's a bug in your script. Redoing the undo (in line
> 674) doesn't work.
> If you want to debug such things, it would be extremely useful to
> copy the gtp-stream (to and from gnugo) into a logfile.


You're right.  I was doing genmove in one case and gg_genmove in the
other.  Here's a patch.

Thanks

Evan Daniel

Index: matcher_check
===================================================================
RCS file: /cvsroot/gnugo/gnugo/interface/gtp_examples/matcher_check,v
retrieving revision 1.16
diff -u -d -r1.16 matcher_check
--- matcher_check       5 Sep 2002 03:26:36 -0000       1.16
+++ matcher_check       12 Sep 2002 13:34:51 -0000
@@ -671,7 +671,7 @@
        #cancel the undo
        my $last_played = "B";
        if ($toplay eq "B") { $last_played = "W"; }
-       print $in "gg_genmove $last_played\n";
+       print $in "genmove_$last_played\n";
        eat_move($out);
     }