Re: [gnugo-devel] regressions and bugs

[Gunnar Farnebäck]

[Moving back to the list.]

Emanuele Cisbani wrote:
> Working on STS_RV, first semeai, first problem:
>
> Class 0 semiai.
> White outside libs= 2
> Black outside libs= 3
> Shared libs=0
> Black wins unconditionally.
> Black to play pass.
> White has two ko threads in either A12, B12, C12.
> Value of the kothread 12 points.
>
>    A B C D E F G H J K L M N O P Q R S T
> 19 . . . . . . . . . . . . . . . . . . . 19
> 18 . . . . . . . . . . . . . . . . . . . 18
> 17 . . . . . . . . . . . . . . . . . . . 17
> 16 . . . + . . . . . + . . . . . + . . . 16
> 15 . . . . . . . . . . . . . . . . . . . 15
> 14 . . . . . . . . . . . . . . . . . . . 14
> 13 O O O O . . . . . . . . . . . . . . . 13
> 12 . . . O . . . . . . . . . . . . . . . 12
> 11 X X X O . . . . . . . . . . . . . . . 11
> 10 O O O X . . . . . + . . . . . + . . . 10
>  9 . X . X . . . . . . . . . . . . . . . 9
>  8 X X X X . . . . . . . . . . . . . . . 8
>  7 . . . . . . . . . . . . . . . . . . . 7
>  6 . . . . . . . . . . . . . . . . . . . 6
>  5 . . . . . . . . . . . . . . . . . . . 5
>  4 . . . + . . . . . + . . . . . + . . . 4
>  3 . . . . . . . . . . . . . . . . . . . 3
>  2 . . . . . . . . . . . . . . . . . . . 2
>  1 . . . . . . . . . . . . . . . . . . . 1
>    A B C D E F G H J K L M N O P Q R S T
>
> restricted_genmove black A9 C9 PASS
> ? invalid coordinate
>
> How can I contruct a valid test without
> using the PASS argument in vertex list?
>
> Thanks.

Restricted_genmove returns PASS if it doesn't like either of the
allowed moves at all. But I think it's better to do these as actual
semeai tests rather than genmove tests.

Here it would look like this:

1 analyze_semeai A11 A10
#? [1 1 (PASS|A9|C9)]
2 analyze_semeai A10 A11
#? [0 0 PASS]

The assumption of "analyze_semeai A11 A10" is that the player with the
A11 stones (black) moves first in a semeai between A11 and A10. The
first number in the result is whether A11 can live; 0 fail, 1 success,
2 success with good ko, 3 success with bad ko. The second number is
whether A10 can be killed, same codes. Finally which move gives the
best result. In this case which is already settled PASS is an allowed
answer but also other moves which bring black further ahead in the
semeai are accepted.

Thus a plain win has codes [1 1] and a plain loss has codes [0 0].
[1 0] means that the player moving first can live but not kill the
opponent, which is either a seki or mutual independent life. The
result [0 1] is not possible as a correct result but GNU Go has been
known to return that occasionally.

A good ko is characterized by the opponent having to make the first
external ko threat whereas a bad ko is the opposite.

For examples of existing semeai tests, see semeai.tst.

/Gunnar