Sender: hmiller@swamp.intercom.net
Original-Sender: hmiller@quake.delmarva.net
To: darn@vxl.imag.net
Subject: Re: New Project! - advice
References: <634cm9$8ma$1@news.ox.ac.uk> <3456BB4B.6378@vxl.imag.net>
From: Howard Miller <hmiller@quake.delmarva.net
In-Reply-To: Darren Grant's message of Tue, 28 Oct 1997 20:27:55 -0800
Newsgroups: rec.games.computer.quake.quake-c
Date: 30 Oct 1997 22:26:29 -0500
Message-ID: <8nu3dy7guy.fsf@swamp.intercom.net>
Lines: 30
X-Newsreader: Gnus v5.3/Emacs 19.34
NNTP-Posting-Host: 204.183.208.24
Path: clanworld.com!news.webspan.net!news1.exit109.com!newsfeed.direct.ca!news1.chicago.iagnet.net!iagnet.net!newsgate.netvigator.com!newsfeed.kcdata.com!news5.kcdata.com!204.183.208.24
Xref: clanworld.com rec.games.computer.quake.quake-c:3780


One possible solution:
You could try writing the genetic code in another programming language (like 
lisp or c or whatever you prefer), then the actual trials for seeing what
method is more successfull could be done using Quake.
If it's possible to have QuakeC (I have no idea) output to a file?  At least
that way, you could have some communication between the Quake program and the
other one doing the more sophisticated code.
You could have some program that creates creatures, and then have them duke it
out in a deathmatch, and the ones who don't get killed get to influence the
next generation....

Hey, if you want any help doing this.....I do have a little experience doing 
something similar (though with grammar and natural language stuff).

As for ai that changes on the fly, it all depends on how well quake-c can be 
integrated with the outside world - with other programs.
Though you do have limitations on memory, you can make quake take a larger 
chunk of memory for itself.

I really know nothing of Quake-c and quake - these are just a few suggestions
and starting points.

Anyone else?






