[gnugo-devel] More testcases

[Gunnar Farneback]

Some more new testcases. The new game record semeai17.sgf was provided
by Jean-Pierre Vesinet and shows GNU Go 3.4 going totally wrong with
semeai in a 9x9 2-stone game and passing when it incorrectly thought
it had lost everything. At the end I've added some extra moves to
point out some more misreadings. In particular the stable final
position is interesting:

   A B C D E F G H J
 9 . O . X O . O O . 9
 8 X X X X O O X O O 8
 7 . O + X O X X X O 7
 6 . . O O O O O X X 6
 5 . O X . O X O O X 5
 4 . X X O O X X X X 4
 3 . . X O X X X . . 3
 2 . . X O O X X O O 2
 1 . . . . O X O O O 1
   A B C D E F G H J

Clearly there's a seki in the lower right but the owl code thinks G8
is alive while H2 is dead. What's really interesting now is that the
semeai code (which isn't called when determining status) has the
opposite opinion, G8 is dead while H2 is alive.

/Gunnar

Index: regression/connection.tst
===================================================================
RCS file: /cvsroot/gnugo/gnugo/regression/connection.tst,v
retrieving revision 1.68
diff -u -r1.68 connection.tst
--- regression/connection.tst   29 Nov 2003 19:33:36 -0000      1.68
+++ regression/connection.tst   23 Jan 2004 13:47:57 -0000
@@ -377,6 +377,11 @@
 109 same_dragon O3 N5
 #? [0]*
 
+# See also nngs4:290
+loadsgf games/nngs/aburry-gnugo-3.3.12-200212162212.sgf 38
+110 connect E3 G3
+#? [0]
+
 # Report number of nodes visited by the tactical reading
 10000 get_reading_node_counter
 #? [0]&
Index: regression/nngs4.tst
===================================================================
RCS file: /cvsroot/gnugo/gnugo/regression/nngs4.tst,v
retrieving revision 1.56
diff -u -r1.56 nngs4.tst
--- regression/nngs4.tst        16 Jan 2004 12:38:58 -0000      1.56
+++ regression/nngs4.tst        23 Jan 2004 13:47:58 -0000
@@ -116,6 +116,7 @@
 280 restricted_genmove black F3 H3
 #? [F3]*
 
+# See also connection:110
 loadsgf games/nngs/aburry-gnugo-3.3.12-200212162212.sgf 38
 290 restricted_genmove black H1 H6 F1
 #? [!H1]*
Index: regression/semeai.tst
===================================================================
RCS file: /cvsroot/gnugo/gnugo/regression/semeai.tst,v
retrieving revision 1.55
diff -u -r1.55 semeai.tst
--- regression/semeai.tst       14 Jan 2004 22:22:37 -0000      1.55
+++ regression/semeai.tst       23 Jan 2004 13:47:58 -0000
@@ -358,27 +358,63 @@
 #? [1 1 (Q19|S19|T18|T16)]
 88 analyze_semeai S18 N19
 #? [2 0 T12]*
+
 loadsgf games/semeai/semeai16.sgf 224
 90 analyze_semeai N19 S18
 #? [1 3 (Q19|S19|T18|T16)]*
 91 analyze_semeai S18 N19
 #? [2 0 Q17]*
+
 loadsgf games/semeai/semeai16.sgf 226
 93 analyze_semeai N19 S18
 #? [1 3 S19]*
 94 analyze_semeai S18 N19
 #? [1 1 S19]
+
 loadsgf games/semeai/semeai16.sgf 230
 96 analyze_semeai N19 S18
 #? [1 0 S19]
 97 analyze_semeai S18 N19
 #? [1 1 S19]*
+
 loadsgf games/nngs/beedee-gnugo-3.5.3-200401140035.sgf 280
 99 analyze_semeai R13 Q13
 #? [1 0 (M12|K9|M6|L7|S17)]
 100 analyze_semeai Q13 R13
 #? [1 1 (P10|Q11|S10|T11)]
 
+loadsgf games/semeai/semeai17.sgf 52
+101 analyze_semeai D9 C6
+#? [1 1 (D6|C7)]
+102 analyze_semeai C6 D9
+#? [1 1 (D6|C7)]
+103 analyze_semeai D9 E9
+#? [1 1 (D6|C7)]
+104 analyze_semeai E9 D9
+#? [1 1 (D6|C7)]
+105 analyze_semeai G8 E9
+#? [1 1 (D6|C7)]
+106 analyze_semeai E9 G8
+#? [1 0 (D6|C7)]
+
+loadsgf games/semeai/semeai17.sgf 60
+108 analyze_semeai G8 H2
+#? [1 1 J2]
+109 analyze_semeai H2 G8
+#? [1 1 E1]
+
+loadsgf games/semeai/semeai17.sgf 62
+110 analyze_semeai G8 H2
+#? [1 0 (PASS|F3|E1)]
+111 analyze_semeai H2 G8
+#? [1 0 (PASS|F3|E1)]
+
+loadsgf games/semeai/semeai17.sgf 64
+112 analyze_semeai G8 H2
+#? [1 0 PASS]
+113 analyze_semeai H2 G8
+#? [1 0 PASS]
+
 ########### semeai gen_move tests #################
 
 # A6 gives an unfavorable ko while F10 gives seki.
@@ -400,6 +436,11 @@
 98 reg_genmove black
 #? [S19]
 
+# Doubtful whether C7 should be accepted even if it kills all white.
+loadsgf games/semeai/semeai17.sgf 52
+107 reg_genmove black
+#? [D6|C7]
+
 ########### semeai status tests #################
 
 loadsgf games/semeai/semeai15.sgf
@@ -415,6 +456,12 @@
 #? [alive]
 86 dragon_status C11
 #? [critical A2 A2]
+
+loadsgf games/semeai/semeai17.sgf 64
+114 dragon_status G8
+#? [alive]
+115 dragon_status H2
+#? [alive]
 
 ########### end of semeai tests #################
 
Index: regression/games/semeai/semeai17.sgf
===================================================================
RCS file: regression/games/semeai/semeai17.sgf
diff -N regression/games/semeai/semeai17.sgf
--- /dev/null   1 Jan 1970 00:00:00 -0000
+++ regression/games/semeai/semeai17.sgf        23 Jan 2004 13:47:58 -0000
@@ -0,0 +1,12 @@
+(;GM[1]FF[3]
+SZ[9]HA[2]KM[5.5]
+PW[Human]
+PB[Computer]
+DT[2003-08-12]
+AB[cg][gc];W[ee];B[ff];W[hh];B[ce];W[ef];B[eg];W[fg];B[dc];W[dg];B[gg]
+;W[eh];B[hf];W[ge];B[gf];W[gd];B[gh];W[gi];B[fe];W[fd];B[fc];W[ec];
+B[hd];W[ed];B[db];W[he];B[hi];W[ii];B[id];W[eb];B[fa];W[fb];B[gb];
+W[ea];B[ie];W[hb];B[hc];W[ib];B[fi];W[hi];B[fh];W[ga];B[if];W[ha];
+B[ch];W[dh];B[eg];W[df];B[cf];W[ic];B[da];W[cd];B[cb];W[dd];B[bb];
+W[bc];B[ab];W[be];B[bf];W[ba];B[tt];W[ih];B[fg];W[ei]
+)