[Shootout-list] I'm not your designated OCaml guy

Brandon J. Van Every vanevery@indiegamedesign.com
Tue, 28 Sep 2004 00:25:21 -0700


Brent Fulgham wrote:
>
> 2.  A new language has been added:  FELIX, by John Skaller.  Brandon
> should have some affinity for this language, since it is largely
> implemented in Objective CAML.  :-)

I am firm on one point: I think C++, Java, and C# all suck and I want
them to die.  The sooner their commercial death, the better.  They are
technologically regressive 'worker cog' languages.  Those jobs are going
to be done best by wage slaves in India et al.  There are no competitive
advantages to knowing them.

However, I don't wish to be pigeonholed as "the OCaml guy."  It's what
led me here, but I'm not feeling quite as keen on it lately.  Tools and
libraries issues on Windows are a really big deal, not to mention the
UNIX-centrism of the OCaml community.  These are difficult things to
change, and I've had little luck trying to galvanize parts of the OCaml
community to change.

Scheme was my designated exit hatch if OCaml didn't work out.  I see
stronger Windows support there, and I may yet play that card.

I've also been interested in John Skaller's FELIX, since unlike the
OCaml community, his business orientation is in line with my own as a
game developer.  But, he doesn't offer MSVC support at present.  That's
presently a showstopper for my Nebula2-driven concerns.  Also the FELIX
community is nil, so that's building an effort from scratch.  Hm, right
culture + no community, vs. wrong culture + large community.  Which is
worse?

Nebula2 has strong Lua support, and possibly acceptable Python support.
In the short term I might choose one or the other out of sheer
pragmatism.

Also I've matured a lot about what my game's AI paradigm is going to be.
The core data structures are going to be RISC-like, in C/ASM.  Then I'm
going to have a pile of ad hoc heuristic scripts on top of it.  Then I'm
going to recombine those using genetic algorithms.  It's possible that
any scripting language would do, or that some languages (like OCaml)
would be better suited to it than others.  I don't know the answer to
that quandry yet.

My current plans are:

- in the short term, get the core C/ASM data structures built, using
whatever language offers the most readily usable support under Nebula2
and Visual Studio 2003.  Currently that's Lua or Python.

- take a stab at implementing an OCaml Script Server for Nebula2, and
see if it actually buys me anything other than headaches.

So, I haven't given up on OCaml, but it's not accurate to say that I'm
"an OCaml guy" right now.  The lack of 'pep' on ocaml-biz is another
contributing factor that has increased my ambivalence.


Cheers,                         www.indiegamedesign.com
Brandon Van Every               Seattle, WA

"We live in a world of very bright people building
crappy software with total shit for tools and process."
                                - Ed McKenzie