[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Bug-gnubg] Extendable rollouts
From: |
Jim Segrave |
Subject: |
Re: [Bug-gnubg] Extendable rollouts |
Date: |
Sun, 13 Jul 2003 23:07:54 +0200 |
User-agent: |
Mutt/1.2.5.1i |
On Sun 13 Jul 2003 (20:27 +0000), Joern Thyssen wrote:
> On Sun, Jul 13, 2003 at 08:31:49PM +0200, Jim Segrave wrote
> > On Sun 13 Jul 2003 (17:18 +0000), Joern Thyssen wrote:
> > >
> > > I've commited a small fix that seems to work for me, althoough I'm not
> > > sure that it's correct.
> >
> > That looks much better - and led me to fix the code which skips over
> > considering rollouts already done which simply wasted time.
>
> err, I can only see a small change in rollout.c: "int" to "unsigned
> int". Was that it?
Yes - I was being lazy and wanted to set a counter to a very big
number and then replace it if any move to be rolled out needed to
start on an earlier game.
int nFirstGame = ~0;
loop over all moves {
if (first game to roll out < nFirstGame)
nFirstGame = first game to roll out
other stuff
}
but ~0 is a big int only for unsigned ints. Sigh.
I'm now doing another little change to text.c = it couldn't tell
Supremo from a user defined setting (the eval context matches World
Class, but the filters are different, the old code only found that the
eval context was World Class, then saw the filters were different and
didn't examine Supremo.
--
Jim Segrave address@hidden