[Bug-gnubg] Backgame test position (was cubeless rollout bug)

[Michaeldepreli]

I have now run this position through GNU at 2-ply (results below) and
it does appear that the GNU 1-ply problem is in evidence here,
unless there's something else I've missed.
Are there plans to eliminate this problem?
Also will this problem occur at all odd plies?
Is the problem more noticeable, the more GNU doesn't understand 
the position?
In my example does it not seem strange that at 1-Ply GNU loses
so many BG's? 
The equity difference compared to other bots is quite
severe, making 1 ply completely unreliable here.
At 1-Ply the wins went up from 31.2 to 39.2 from the old build.
Was this because of the previous bug?
What exactly was fixed (sorry I missed it in the mailing list).

32000 trials at 0-Ply:

JF              0-Ply    -0.672    40.2/0.9/0.0---59.8/35.0/13.6
SW3.2           0-Ply    -0.669    40.4/1.0/0.0---59.6/35.7/13.0
SW4             0-Ply    -0.636    42.1/1.0/0.0---57.9/35.7/13.1
GNU             0-Ply    -0.590    43.5/0.9/0.0---56.4/34.7/11.5

2160 trials at 1-Ply (13 moves at 0.308)

GNU (Old Build) 1-Ply    -0.947    31.2/0.9/0.0---68.8/40.6/17.4
GNU (New Build) 1-Ply    -0.781    39.2/0.6/0.0---60.8/40.4/16.7
JF              1-Ply    -0.607    42.4/1.1/0.0---57.6/33.6/12.9
SW3.2           1-Ply    -0.545    44.1/1.6/0.1---55.9/34.0/10.3
SW4             1-Ply    -0.552    44.8/1.1/0.0---55.2/33.6/12.3

2160 trials at 2-Ply 50% speed (13 moves at 0.308---7 moves at 0.154) 
                                       
GNU        2-Ply        -0.546  45.3/0.9/0.1---54.7/34.1/12.1
SW3.2      2-Ply        -0.498  45.7/1.4/0.1---54.3/32.7/9.9
SW4        2-Ply        Not yet rolled out

Thanks