[gnugo-devel] 3.3.6 and cvs differences

[Evan Berggren Daniel]

The following is the result of replaying
gnugo-3.3.6-viking4-200208280823.sgf (in my recent regression patch) with
current cvs as test.  Interesting is that there is a long sequence where
the two would have played differently (the game follows the stable
version), from moves 8-20.  Only two moves out of the whole game (300)
differ outside that.  Perhaps this is useful for tuning?

Thanks

Evan Daniel

random seed = 1030923937
At move 8, white:
Test version played   H3
Stable version played D13
At move 9, black:
Test version played   F4
Stable version played O4
At move 10, white:
Test version played   F4
Stable version played O17
At move 11, black:
Test version played   F4
Stable version played M17
At move 12, white:
Test version played   F4
Stable version played R17
At move 13, black:
Test version played   F4
Stable version played O4
At move 14, white:
Test version played   F4
Stable version played Q17
At move 16, white:
Test version played   F4
Stable version played O18
At move 17, black:
Test version played   F4
Stable version played O4
At move 18, white:
Test version played   F4
Stable version played L19
At move 19, black:
Test version played   F4
Stable version played O4
At move 20, white:
Test version played   O7
Stable version played H2
At move 27, black:
Test version played   G6
Stable version played R13
At move 82, white:
Test version played   E7
Stable version played E6