[gnugo-devel] usefull but inefficient pattern

[max-d]

Pattern A6xx

|.OY
|.X*
+---

:8,s

|.aY
|.X*
+---

;lib(a)>1

This pattern is needed to solve such problems (which seem to be frequent )

(;RU[Japanese]SZ[19]HA[0]KM[5.5]
;W[om]
;W[pk]
;W[po]
;W[pq]
;W[qk]
;B[ql]
;B[qm]
;W[qo]
;W[qq]
;W[rk]
;B[rl]
;W[rm]
;B[rn]
;B[ro]
;B[rp]
;B[rq]
;W[rr]
;B[sl]
;W[sm]
;W[sn]
;W[sp]
;W[bq]
;W[cr]
;W[bs]
;B[ap]
;B[bp]
;B[cp]
;W[cq]
;B[dq]
;B[dr]
;W[ar]
;B[aq]
;W[sq]
;B[so]
;W[sm]
;B[sr]
)

OWL status correctly reports the S6 group as critical but  unfortunately,
Gnugo has no care of that and plays elsewhere (even using nofuseki option).
if i put the pattern in pattern.db then Gnugo _does_not_ understand the
position but plays the pattern's move .
(I wonder if the ;lib(a)>1 constraint is usefull here or has just for (bad)
effect to slow down the engine) .


This problem is the followup of this one .
here if we try BT2 instead of BS1(gnugo with the given pattern do so)  , OWL
never even tries WT1.
(the left lower corner is just giving a ko threat).

(;RU[Japanese]SZ[19]HA[0]KM[5.5]
;W[om]
;W[pk]
;W[po]
;W[pq]
;W[qk]
;B[ql]
;B[qm]
;W[qo]
;W[qq]
;W[rk]
;B[rl]
;W[rm]
;B[rn]
;B[ro]
;B[rp]
;B[rq]
;W[rr]
;B[sl]
;W[sm]
;W[sn]
;W[sp]
;W[bq]
;W[cr]
;W[bs]
;B[ap]
;B[bp]
;B[cp]
;W[cq]
;B[dq]
;B[dr]
;W[ar]
;B[aq]
;W[sq]
)