Re: [gnugo-devel] Question about scoring and endgame

[Daniel Bump]

I wrote:

>> I was sent the following 9x9 game which results in an infinite recursion 
>> in (owl.c) do_owl_analyze_semeai if you ask gnugo to load the file, and 
>> then play for white (genmove_white).
>
>This should be fixed by implementing a komaster scheme in
>do_owl_analyze_semeai. I'll try to get this fixed before
>3.1.24.

Actually a komaster scheme won't help here. 

As far as I know the problem of double ko is reasonably
under control for the tactical reading code but not for
the relatively new semeai code, this example shows.

In any case we'll have to fix this before 3.1.24. At
the moment, loading this file causes a segmentation
fault.

This position is actually a triple ko and a good example
for the reading code, too.

$ gnugo -l infinity.sgf --quiet -t --semeai-variations 100 -\
-decide-string J3 -o vars.sgf
J3 can be attacked with ko (good) at H6 (10808 variations)
J3 can be defended at F5 (4 variations)

This is probably more variations than necessary.

Dan