RE: [gnugo-devel] tuning patch

[Portela Fernando]

Paul wrote:

> it seems logical, but doesn't work unfortunately. note this comment:
> 
>  # tm New Pattern (3.1.19)
>  #   s classification, in case of ko (see trevorc:180)
>
> in case of ko oplay_disconnect(*,b,*) returns true, because connection
> code claims victory if one of the strings is captured (even if the
> capture is one stone in ko).

I'm not very familiar with the connection code, but it seems to me that
it should return KO_(something) instead (I'm not convinced that it is
not already the case, but you're maybe better informed than I am)

With a correct return code, we could then use

 ;proper_eye(a) && oplay_disconnect(*,b,*) != WIN

/nando