Re: [gnugo-devel] Tiny boards

[Gunnar Farnebäck]

John wrote:
> I can see 0x0 being a bit questionable, but 1x1 makes a good
> test-case if nothing else.

If you insist...
The patch below reduces MIN_BOARD to 1 and avoids to resign
immediately on 1x1 and 2x2.

> And if GnuGo is to allow non-square boards, then 1x2 offers some
> opportunities for fun play and testing of ko rule implementations...

Non-square boards would require a fair bit of infrastructural
upgrades. It's much, much easier now than it was before we changed to
1D board representation, but still some work. I don't feel it's worth
the trouble at this time but if someone does it would probably get
accepted.

/Gunnar

Index: engine/board.h
===================================================================
RCS file: /cvsroot/gnugo/gnugo/engine/board.h,v
retrieving revision 1.19
diff -u -r1.19 board.h
--- engine/board.h      7 Dec 2004 04:31:05 -0000       1.19
+++ engine/board.h      15 Dec 2004 03:25:51 -0000
@@ -72,7 +72,7 @@
 
 #define DEFAULT_BOARD_SIZE 19
 
-#define MIN_BOARD          2       /* Minimum supported board size.   */
+#define MIN_BOARD          1       /* Minimum supported board size.   */
 #define MAX_BOARD         19       /* Maximum supported board size.   */
 #define MAX_HANDICAP       9       /* Maximum supported handicap.     */
 #define MAX_MOVE_HISTORY 500       /* Max number of moves remembered. */
Index: engine/genmove.c
===================================================================
RCS file: /cvsroot/gnugo/gnugo/engine/genmove.c,v
retrieving revision 1.100
diff -u -r1.100 genmove.c
--- engine/genmove.c    7 Dec 2004 04:50:02 -0000       1.100
+++ engine/genmove.c    15 Dec 2004 03:25:52 -0000
@@ -751,7 +748,7 @@
    * it looks like all our dragons are dead and the generated move
    * is a pass.
    */
-  if (move == PASS_MOVE && !lively_dragon_exists(color))
+  if (board_size > 2 && move == PASS_MOVE && !lively_dragon_exists(color))
     return 1;
   
   if (move == PASS_MOVE