Re: [gnugo-devel] final patch, second version

[Arend Bayer]

Trying out this patch I noticed that --replay -o .. -O v is broken
(this has nothing to do with Paul's patches) with respect to
reevaluationg ko threaths and devalueing of blunders. While the latter
is my fault (arend_3_7.5), the first has probably been around for a
while.

Should probably be fixed by having review_move_reasons() explicitly export
all move values to the global variable potential_moves[][] at its very end
(instead of the somewhat magic call to move_considered() at the end of
value_move_reasons()).

Arend