Re: [gnugo-devel] regress 13x13b:27

[Evan Berggren Daniel]

This is indeed fairly simple.  CD106 can probably be completely removed; I
doubt it is neccessary any more.  This was probably added before the owl
code was very good, and was needed to prevent GNU Go from dying in some
cases.  You could try running the full regression suit without it and see
whether it is still needed elsewhere; if it is, then solving those
problems and sending one collected fix would be a good thing.  But
whatever the results, if you could send them to us along with the relevant
patch, that would be best.

Evan Daniel

On Sun, 6 Apr 2003, kevin yong wrote:

> Hi, Evan:
>
>   i did some work on 13x13b:27.
>
>   problem: B2 wrong, C4 right.
>
>   this is a relatively simple case. i found the source
> of problem:
>
>   originally, C4 got higher points (25.75) than B2
> (23.01). However, because of pattern CD106 matched for
> B2, which calls replace C4 with B2. So, points
> redistributed from C4 to B2, B2 ends 25.75, C4 ends 0.
>
>   solution would just somehow enhance pattern CD106. i
> like to hear your comments first.
>
>   thank you again for all your replies.
>
> kevin.
>
>
> ______________________________________________________________________
> Post your free ad now! http://personals.yahoo.ca
>
>
> _______________________________________________
> gnugo-devel mailing list
> address@hidden
> http://mail.gnu.org/mailman/listinfo/gnugo-devel
>
>