Re: [gnugo-devel] Inessential worms

[Gunnar Farneback]

Dan wrote:
> Then it would seem that readconnect is going to have to
> be able to either consult the semeai module or duplicate
> its functionality. If the potentially inessential worm
> has a lot of liberties the tactical reading code won't
> be able to determine whether or not it is truly inessential.

Saying that readconnect should get the responsibility for amalgamation
was an oversimplification. To be more precise the amalgamation should
be handled by some piece of code which primarily uses readconnect to
decide whether or not to amalgamate two worms. It could also call a
semeai module if necessary, but I'm not certain it would have to.

> Could you explain what you envision readconnect doing in
> a situation like semeai6.sgf (upper left)?

I don't see that it has to do anything at all. There's no need to
amalgamate anything in that area. Even if the semeai had already been
settled I don't see any need to do amalgamation

/Gunnar