[gnugo-devel] regression of recent tuning patches

[Paul Pogonyshev]

I first ran the complete regressions and then separately
verified the source of each change.

I also revised the owl tuning patch a bit: it is useless
to specify which dragon runs to which as connection works
in both directions.  So I used

    owl_escape_value(A) > 0 || owl_escape_value(B) > 0

instead of

    owl_goal_dragon(A) && owl_escape_value(B) > 0

as other similar patterns do.  Revised patch is in the end.


Full breakage:

ego:8           PASS F12 [F12]
nngs:510        FAIL P13 [G14]
handtalk:15     PASS J3 [J3]
auto04:2        FAIL 0 [!0]
ninestones:150  PASS C5 [C5|D5]
kgs:130         PASS S18 [S18]
kgs:490         PASS H13 [H13]
kgs:500         PASS O15 [O15]

6 PASS, 2 FAIL


Eye patterns tuning breakage:

kgs:130         PASS S18 [S18]

  The targeted test case, solved properly.


Atari-atari pattern breakage:

handtalk:15     PASS J3 [J3]

  Good, solved.

kgs:490         PASS H13 [H13]

  The targeted test case, solved properly.


Owl tuning breakage:

ego:8           PASS F12 [F12]

  Did not verify.

nngs:510        FAIL P13 [G14]

  Don't understand it, probably accidental.

auto04:2        FAIL 0 [!0]

  Mostly accidental.  Valid move doesn't falls off because of
  branching cut off: new D1393 has a higher value.  But the
  reading is not quite perfect anyway.  This is a case where
  it is actually ``getting deep, looks lively'', but GNU Go
  doesn't think so.  More often it is the way round.

ninestones:150  PASS C5 [C5|D5]

  Accidental: owl node limit happens in a different reading
  branch now in reading of G15.

kgs:500         PASS O15 [O15]

  The targeted test case, solved properly.


Paul



--- owl_defendpats.db   05 Apr 2005 21:04:14 +0300      1.129
+++ owl_defendpats.db   08 Apr 2005 00:30:41 +0300      
@@ -7022,6 +7022,39 @@ O.X
 ;lib(A) == 2
 
 
+Pattern D1393
+# pp New pattern, see kgs:500 (3.7.4)
+
+O.O.       secure escape connection by first threatening opponent
+?*XO
+
+:8,B,value(50)
+
+A.Ca
+?*DB
+
+# Note: xplay_disconnect() to avoid generating this move when it is
+# not really needed.
+
+; (owl_escape_value(A) > 0 || owl_escape_value(B) > 0)
+; && xplay_disconnect(A,C) && oplay_attack(*,a,D)
+
+
+Pattern D1394
+#pp New pattern, see kgs:500 (3.7.4)
+
+O.O*       connect out!
+??XO
+
+:8,B,value(70)
+
+A.O*
+??XB
+
+; (owl_escape_value(A) > 0 || owl_escape_value(B) > 0)
+; && safe_xmove(*) && !oplay_disconnect(*,A,B)
+
+
 #########################################################
 #                                                       #
 #                          Ko                           #