[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [gnugo-devel] Current CVS very broken

From: Gunnar Farneback
Subject: Re: [gnugo-devel] Current CVS very broken
Date: Mon, 29 Oct 2001 20:02:06 +0100
User-agent: EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/20.7 (sparc-sun-solaris2.7) (with unibyte mode)

Trevor wrote:
> BTW, here's a URL w/ the HTML representation of the current
> regression test, as of this morning:
> (This is still a bit of a work in progress, so I've left the old, stable
> link w/ the 3.1.10 regressions.)

Minor wish list item: Total sums of the number of passes, unexpected
passes, and so on.

> The "x" in the lower-left corner of the second diagram should be a "."?
> Pattern VA19
> # gf New pattern. (3.1.11)
> OX.x       peep at
> edge
> X.*.
> XX..
> ----
> :8,s,value(75)
> OX.x
> A.*.
> AA.x
> ----
> ;lib(A)==2 &&
> owl_big_eyespace(*,A)

Yes, I think that's what I intended.

> +Pattern VA20
> +
> +?X?
> +X.X
> +*X?
> +X.X
> +?X?
> +
> +:8,-,value(75)
> +
> +?X?
> +XaX
> +*C?
> +XbX
> +?X?
> +
> +; owl_topological_eye(a,C) == 2 && 
> +; owl_topological_eye(b,C) == 2

One thing to be aware of is that the owl_topological_eye autohelper is
computationally expensive, so it should only be used where it's really
needed. In this pattern it looks well motivated.

>  - It seems that a lot of the owl problems depend on connections, and
> there are a number of connection-like patterns w/in the owl patterns.
> Is this the long term idea, or will the OWL code eventually all the
> connection code directly to determine cuts/connections?  This new
> test case is a perfect example of a very simple cut to kill.

To add to Dan's answer, there are two issues with connections in the
owl code. The first one is that the owl code has very little
understanding of the possibility that the dragon becomes cut into
pieces. There are examples where it gladly builds two eyes but where
those end up in disconnected parts of what was a single dragon to
start with. The second issue is with connections to external,
preferrably living, dragons. This works better but does indeed require
quite a few patterns.

My hope is that the next generation of the owl code will be able to
make use of a reading connection analyzer, allowing a better
understanding of internal connections and requiring fewer patterns to
deal with external connections.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]