Re: [gnugo-devel] Tests

[bump]

I've added the new tests in ninestones.tst to the CVS taking Gunnar's 
comments into account.

> > +loadsgf games/nngs/leftd-gnugo-3.3.16-200302072009.sgf 76
> > +360 gg_genmove black
> > +#? [!S3]
> 
> Same comment again. S4 is locally better than S3, I assume.

I think S5 is the right move locally.

(;GM[1]FF[3]
SZ[19]
GN[GNU Go 3.3.16 load and print]
DT[2003-02-12]
KM[-0.5]AP[GNU Go 3.3.16]RU[Japanese];
AW[lb][mb][mc][qc][rc][ld][nd][qd][le][ne][oe]
[pe][lf][nf][qf][rf][og][ch][oh][gm][gn][fo]
[go][mo][no][po][ep][hp][op][cq][dq][eq][fq][lq][pq][qq][or]
AB[ma][nb][lc][nc][oc][dd][gd][jd][md][od][pd][rd]
[qe][re][of][pf][mg][ng][pg][ph][oi][dj][jj][pj]
[em][fm][jm][pm][dn][fn][mn][nn][on][do][eo][ho]
[lo][oo][cp][dp][fp][gp][jp][pp][qp][qr]
PL[B]
(;B[rp];W[rq]C[Sente since if B does nothing ...
];B[tt];W[qn];B[qo];
W[sp];B[pn];W[ro];B[po];W[rm];B[qm];W[rn];B[rl]
C[... W gouges out the bottom.
])
(;B[ro]C[Looks like gote but ...
];W[rp];B[qo]
C[B is left with a big sente yose move at S3.
]))

> > +# B can win the semeai. It takes 11 moves to remove the W
> > +# stones from the board, including going back to capture
> > +# D18. It takes 11 moves to remove the B stones from the
> > +# board. The game move at L9 fills one of B's own liberties.
> 
> In my vocabulary a move like G5 would fill one of B's own liberties.
> L9 plays at a liberty while creating a new one and is with respect to
> the liberty balance equivalent to a tenuki.

Technically you are right, but the L9 M9 exchange results in
B losing one liberty in a semeai where the liberty count is
close. It does not fill a liberty but forces W to fill it.

Yes, it is probably true that B can barely win the semeai.
W can try W:A8 B:B5 before filling at M9 and still B is one
move ahead. Actually there is a ko but no ko threats.

I think when a semeai is this close the policy should be
to start playing it. 

Dan