Re: [gnugo-devel] attack1() patch

[Evan Berggren Daniel]

> So how does this fourth case look? You never showed us.

Essentially, it is case A, but with the outside stones owl attackable.
So, something along these lines (assume all outside stones invincible):

Case D:

+-----
|.X.XO
|XXXOO
|OOOX.
|O..X.
|O..X.
|O..X.
|XXXX.

The lower O stones are only tactically attackable at reading depths <
fourlib_depth.  However, they are owl attackable at much greater depths.
This is also the case if the O stones have only 3 liberties at greater
reading depths.  Here, the lone X stone should not really be tactically
dead, but it is probably ok to treat it as such (as you suggested, given
that defender must win the ko several times).  Other similar cases might
be where the O stones had fewer direct liberties, but a capture required a
more complex move that is not looked for at greater reading depths.
However, the defender would still have to win the ko several times.

Thanks

Evan Daniel