[Top][All Lists]

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

RE: Work in progress (was : [gnugo-devel] a patch from a newbie : ))

From: Portela Fernand
Subject: RE: Work in progress (was : [gnugo-devel] a patch from a newbie : ))
Date: Mon, 16 Sep 2002 01:59:45 +0200

Arend wrote:

> >       owl_attack_move_reason_known()
> >       owl_defense_move_reason_known()
> I am not sure you should add code here. These are mostly used to e.g.
> forget the strategic effect of a STRATEGIC_DEFEND_MOVE if we have
> already counted the move as an owl defense (otherwise we would count the
> effect twice). But if it's a OWL_DEFENSE_MOVE_LOSS, we should probably
> still count the strategic effect.

Ok, I'll make these optional/deactivated with #if 0 directives.

> >       estimate_territorial_value()
> Just to understand what you are doing: Did you explicitly add the
> effective worm size to the territorial value? (The other option would be
> to have the worm marked as critical/captured/defended for evaluation by
> influence.c).

I'm using mark_changed_dragon(), which I modified a bit. So, the worm just
gets marked. This way seemed natural to me, maybe I'm wrong.

> Well, I cannot speak for Dan and Gunnar, but it looks like a clear
> improvement, so I'd just say go ahead with it.

Well, I already did the "dirty" job I was talking about while waiting for
regression tests to finish. :)

> I don't expect you have to touch that. (Depending on details of your
> patch I could imagine adding a case there, though.)

Ok, we'll see.

> I always think getting some PASSes by doing right is worth more
> than PASSes by (say) some tuning that could turn out badly in other
> situations :)



reply via email to

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