bug-gnubg
[Top][All Lists]
Advanced

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

Re: [Bug-gnubg] Re: The importance of METs


From: Jim Segrave
Subject: Re: [Bug-gnubg] Re: The importance of METs
Date: Tue, 2 Sep 2003 11:09:20 +0200
User-agent: Mutt/1.4i

On Tue 02 Sep 2003 (02:32 +0200), address@hidden wrote:
> On Mon, 1 Sep 2003, Jim Segrave wrote:
> 
> > Start a copy of gnubg, you may as well use the nogui version. This
> > will be the master.
> > Enter the command
> >
> >    external 127.0.0.1:4000
> >
> > where 127.0.0.1 means the local host, use port 4000
> >
> > Start another copy of gnubg.
> > Go to Settings->Players. At the bottom, click external and fill in
> > 127.0.0.1:4000
> >
> > Start a match/session whatever and watch it play.
> >
> > I believe, from looking at the code, that the master server will be
> > playing using the Evaluation settings for cube and chequer play, not
> > the player 0 settings.
> 
> I have run  into a problem: the "master" seems to  be making random cube
> decisions. I've set everything  to "expert" level including "evaluation"
> and load  the .gnubgautorc both in  master and client.  Yet the "master"
> takes doubles when closed out and nonsense like that.

can you make an .sgf file of one of these and send it to me - I don't
need the analysis. The only obvious thing I can see from a *very*
quick skim of the code is that it will take when FindCubeDecision()
returns any of:

NODOUBLE_DEADCUBE
NO_REDOUBLE_DEADCUBE
OPTIONAL_DOUBLE_BEAVER
OPTIONAL_DOUBLE_PASS
OPTIONAL_REDOUBLE_PASS

Are any of these likely to have been the state (as viewed by the
master?).

Otherwise the code in external.c:CommandExternal() is incorrect in
some other way - this is *not* the same code as in play.c which is
normally used by gnubg, so it wouldn't surprise me if there may be
some other more fundamental bugs here, since I doubt it's been used
much, if at all. There might be simple silly errors like not switching
sides or setting the player on roll or something when doing the cube
decisions.


-- 
Jim Segrave           address@hidden




reply via email to

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