Re: [gnugo-devel] paul_3_13.5.gz

[bump]

Paul wrote:

> if no one has any objections, i think we can safely add the patch
> to cvs. it doesn't break anything in regression, so it must be
> bug-free. and i already know how to make it faster than the
> current implementation (even without much work). it might in
> principal break some parts of code which are not used currently
> (like tree pattern matcher) but they will be easy to fix as soon
> as anybody ressurect them.

I would not necessarily say that something is not in use
if it is not configured by default.

This patch breaks compilation after configure --enable-oracle.

I think the following bit should take care of that. Let me
know if it looks incorrect.

It also breaks --enable-experimental-reading I haven't tried
to fix that.

Dan



Index: engine/oracle.c
===================================================================
RCS file: /cvsroot/gnugo/gnugo/engine/oracle.c,v
retrieving revision 1.7
diff -u -r1.7 oracle.c
--- engine/oracle.c     21 Nov 2002 00:25:10 -0000      1.7
+++ engine/oracle.c     6 Dec 2002 18:04:46 -0000
@@ -81,7 +81,7 @@
   const char *reason; /* why this move */
 };
 
-static void oracle_callback(int m, int n, int color, struct pattern *pattern,
+static void oracle_callback(int anchor, int color, struct pattern *pattern,
                            int ll, void *data);
 static void oracle_add_move(struct oracle_move_data *moves, 
                            int this_move, int this_value, 
@@ -294,14 +294,14 @@
 }
 
 static void
-oracle_callback(int m, int n, int color, struct pattern *pattern,
+oracle_callback(int anchor, int color, struct pattern *pattern,
                int ll, void *data)
 {
   int this_move;
   struct oracle_move_data *moves = data;
   UNUSED(color);
 
-  this_move = AFFINE_TRANSFORM(pattern->movei, pattern->movej, ll, m, n);
+  this_move = AFFINE_TRANSFORM(pattern->move_offset, ll, anchor);
   if (within_search_area(this_move))
     oracle_add_move(moves, this_move, pattern->value, pattern->name);
   else