[gnugo-devel] An interesting regression case

[SP LEE]

I'm not sure if gnugo got the whole idea, but it happens to be right in the 
regression golife:7.

# See also test case owl:192
# F3 does kill black.
#
#    A B C D E F G H J
#  9 . . . . . . X O . 9          O=white
#  8 X X . O . O X X O 8          X=black
#  7 O X O . . O O O . 7
#  6 O O O X X X O . O 6
#  5 . . X . X O . O X 5
#  4 . . X 3 4 X O O . 4
#  3 O . X . 2 1 O . . 3
#  2 . O . X X X X O . 2
#  1 . . . 7 . 6 5 . . 1
#    A B C D E F G H J
#
loadsgf games/golife.sgf 46
7 gg_genmove white
#? [F3]*

Because of the complication of the corner, black can't be killed by W:F3. 
For example, W:F3, B:E4, W:G1, B:C1, W:C2, B:A4, W:B4, B:B3, W:A5, B:A2, 
W:A1, B:D1, where W has to let B to make an eye at E1. Gnugo does see the 
good W move at A4, but maybe missing the details thereafter.

Therefore, the following patch removes this test case.

SP Lee

Index: golife.tst
===================================================================
RCS file: /cvsroot/gnugo/gnugo/regression/golife.tst,v
retrieving revision 1.22
diff -u -r1.22 golife.tst
--- golife.tst  22 Jun 2003 17:12:12 -0000      1.22
+++ golife.tst  1 Jul 2003 03:52:23 -0000
@@ -45,9 +45,11 @@
#  1 . . . 7 . 6 5 . . 1
#    A B C D E F G H J
#
-loadsgf games/golife.sgf 46
-7 gg_genmove white
-#? [F3]*
+#Wrong defense sequence for W. One of the good sequences is:
+#W:F3, B:E4, W:G1, B:C1, W:C2, B:A4, W:B4, B:B3, W:A5, B:A2, W:A1, B:D1
+#loadsgf games/golife.sgf 46
+#7 gg_genmove white
+##? [F3]*

# Connecting the one point ko at G5 is too small. Either C2 or G1 looks
# largest.

_________________________________________________________________
Add photos to your messages with MSN 8. Get 2 months FREE*.  
http://join.msn.com/?page=features/featuredemail