Re: [gnugo-devel] tuning patch

[Arend Bayer]

Martin wrote:

> After applying my last patch
> (http://mail.gnu.org/archive/html/gnugo-devel/2004-01/msg00120.html),
> this pattern gives regression delta
>
> ld_owl:406      PASS 1 R3 [1 R3]
> ld_owl:407      PASS 0 [0]
> ld_owl:408      PASS 1 T3 [1 T3]
> ld_owl:409      PASS 0 [0]
> ld_owl:411      PASS 0 [0]
> ld_owl:412      PASS 1 T2 [1 (R3|T2)]

Nice result!

> Furthermore, I suspect that the move R3 which is now generated for
> ld_owl:410 is valid, but the test requires T2. Could someone
> double-check this?

Yes, that is right. (If you get "1 R3", that is.) Btw, R3 in ld_owl:412
doesn't make sense (illegal move) so probably someone added R3 to the
wrong test.

> @@ -5747,6 +5747,20 @@
>  ;lib(A)==3
>
>
> +Pattern A1351
> +# mh New Pattern (3.5.4)
> +# see ld_owl:406-412
> +
> +|OO??   better than capturing
> +|XOX?
> +|.X.X
> +|*.X?
> +|..X?
> ++----
> +
> +:8,s,value(75)
> +
> +

One might try to make it more general, but nothing wrong with adding it
as it is, of course.

Arend