[Bug-gnubg] Problems with latest Windows Build

[Ned Cross]

Build 0.14 Devel 2004Oct28 on Win2000.

1) I have to use the English-UK setting because the English-US puts a huge
whitespace in the analysis window making it unusable (I think this was
already identified, and it's not a big deal to use English-UK but some users
might not think of that).

2) Settings/Analysis will not let me do full cube nor chequer analysis
"without" using neural net pruning.  Same problem for Evaluation settings as
well.  --> Do settings/analysis and UN-check "use neural net pruning", OK
and then settings/save settings.  "Settings saved to .gnubgautorc", yet
close and reopen GNUBG and settings/analysis reveals "use neural net
pruning" is checked again! (likewise for Evaluation).

3) Move filter for Tiny is still selecting 8 candidates, 0.080.  Seems like
a lot of candidates for "tiny".  I recall some users doing rollout tests
that gave very good results with 5 candidates, .100.  OK nevermind I just
checked the 29SEP2004 buld and the tiny filter was 8 candidates then as well
so I guess that's how it has been...I thought it used to be 4 candidates.

4) Settings/Analysis GUI: the tooltips on the analysis and evaluation
settings dialogs sometimes persist instead of disappearing.  This makes it
difficult to make selections, and I sometimes have to drag the window across
the screen to uncover the OK button so I can close the dialog and continue
using the program.  This has been going on for some time, not just with this
latest build, but I suppose it could have something to do with my graphics
card if noone else is having the trouble.

5) Of course the biggest problem of all is the sometimes poor results of the
pruning net in rollouts combined with the inability to use the old
evaluations at reduced speed.  2-ply rollouts performed at 50% checker and
33% cube speed have proven quite accurate and considerably faster than 2-ply
100%.  2-ply prune rollouts, while faster per number of trials, are not
proving as accurate, and have very high standard error rates, making them
possibly unusable.