I was jsut looking at the fear projuect site, and it doesn''t look like there is too much new from the last time I downloaded fear. (Around four or five months ago). You cooking up anything new? Any great new ideas for it? Have you thought about creating a parrelel processer that could be used with fear? I think that would be pretty cool. BTW, I love your book, I read as much of it as I could on a two hour visit to B&N, but taking it home with me was a little out of my price range.... My Kung Fu is stronger. neo88

There''s been a whole lot of new stuff added recently. The 0.3 design established a solid base for FEAR modules, and improved the modelling of AI architectures. I expect the 0.3.x branch to go on for quite a bit, adding various little features that don''t break the API or change the modelling language. I''ll be polishing up existing world interfaces and some AI techniques too, as well as adding to the documentation -- which is already shaping up nicely.


I''m happy with FEAR 0.3; it does what it''s supposed to do, but there are some major milestones ahead in my ultimate quest:

• How to establish relations between formal interfaces. For example how do you express that binary (or categorical) pattern recognition is a subset of continuous PR? If you can express that, you can interchange components with similar functionality (e.g. swap DT and NN with a single click).
  
• Can you create a simple object system without building an entire game logic? How do you do it so it is flexible enough to handle complex worlds and non-trivial NPC (i.e. so it doesn''t rely on a clumbsy hierarchy of static C++ classes)?
  
• How do you reuse common design patterns on a high-level (like subsumption architectures or voting systems) and is it possible to define them with interchangeable behaviors?
  

I have the nagging feeling that one awesome design can solve all these problems in one go... now if I can only get my head around it

Alex

---

AiGameDev.com

Yeah, I think you are right. Have you ever thought of creating your own programming language for this kind of stuff. Cause C++ is severlely limited for some AI applications. I have been thinking about creating a language that would specilize in the creation of AI agents that could prefrom any task that the implementer wishes.

My Kung Fu is stronger.
neo88

heh, you''re close Though there''s little need for another language... I think I''ll be embracing the lisp way, but that''s as much as I''ll say at this early stage.

Alex

quote:

---

Original post by neo88
Yeah, I think you are right. Have you ever thought of creating your own programming language for this kind of stuff. Cause C++ is severlely limited for some AI applications.

---

That''s what Lisp is for. Haven''t you been reading the "Why Lisp?" thread?

hee hee,
Predictor
http://will.dwinnell.com

Indeed Predictor, but my attempt to start a "Lisp, How?" thread has fallen flat on its face...

lol. Actually, I haven''t started to read the why lisp? thread yet I''ll do that right know. hehe. I don''t know to much about lisp yet so maybe I should download the complier too. I will say that I am working on a programming language that fufills the requriments mentioned above, I will disscuss implementation with any of you who want too through email.

My Kung Fu is stronger.
neo88

neo88, those aren''t really problems that can be solved with a language. It''s a matter of working out a design, then implementing it. It would be possible in C++ -- with a lot more work than lisp.

I''ll probably use scheme to test out the design, but it will end up supporting C++ too (just because I have a lot of C++ code and it''s an industry standard).