Complete.Org: Mailing Lists: Archives: freeciv-dev: January 2000:
[Freeciv-Dev] Re: Why client-side AI could be a Bad Thing
Home

[Freeciv-Dev] Re: Why client-side AI could be a Bad Thing

[Top] [All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index] [Thread Index]
To: Jules Bean <jmlb2@xxxxxxxxxxxxxxxx>
Cc: freeciv-dev@xxxxxxxxxxx
Subject: [Freeciv-Dev] Re: Why client-side AI could be a Bad Thing
From: Arkadiusz Danilecki <szopen@xxxxxxxxxxxxxxxxxxxx>
Date: Fri, 28 Jan 2000 20:49:49 +0100 (MET)

On Fri, 28 Jan 2000, Jules Bean wrote:

> On Fri, 28 Jan 2000, Arkadiusz Danilecki wrote:
> 
> > 
> > On Fri, 28 Jan 2000, Jules Bean wrote:
> > 
> > A.D.Danilecki wrote: 
> > > > >1) AI should cheat. Yes, it should. Unless someone create
> > 
> > > I don't like cheating.
> > 
> >     I don't like cheating too, but i like long hard games, and it is
> > hard for such a game with freciv AI
> 
> FreeCiv is a positive feedback game.  So you don't really get 'long hard'
> games.  Games are only hard until you have the upper hand, but once you
> get the clear upper hand, you are more or less unbeatable.

Well... with ten hard AIs you first struggle to survive, then you have
pleasure of defeating first AI and them real fun of crushing all AI with
using of modern weapons :)

> > > Instead, use a handicap.
> > 
> >     Handicap can's be used instead. If ai is stupid, it will play
> > stupid no matter how big advantage it gets at beginning. When it can
> > cheat, it can at least try be equal opponent to human players.
> >     When i think about cheating, i think more about knowledge about
> > map etc. not breaking all rules.
> 
> I don't really see this.  I'm sure a handicap will do.  Double its hit
> points of all its units, say.

        Hm... 
[cut]
> > i can agree with this. So, instead of having two processes on my poor k166
> > i will have 16 processes. Hm. No overhead right?
> 
> *sigh*
> 
> No.
> 
> This is one of the most frequently misunderstood things about Linux and
> unix-style systems.  Systems with an advance VM system and advanced
> multitasking system suffer a neglible overhead for splitting a complex
> task into several processes in a logical fashion.
> 
> I'm not saying there is *no* overhead, but if we look to improve freeciv
> performance, simply improve the goto functions :-)

        Well... 

        As far as i understand, switching context between processes on
weak, heavy loaded machines _is_ expensive. I have such a machine. 
        
        But after reading this and other posts i must rethink my thoughts
about server-side AI

A.D.Danilecki "szopen"


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