From: "netnews.hinet.net" <chris AT NOSPAM DOT beyoung DOT evercom DOT com DOT tw>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: The future of graphics programming
Date: Fri, 13 Mar 1998 13:14:45 +0800
Organization: DCI HiNet
Lines: 49
Message-ID: <6eaevj$3vt@netnews.hinet.net>
References: <6cvn02$6t5 AT netnews DOT hinet DOT net> <34faed92 DOT 21378329 AT news DOT eunet DOT be> <6dugbb$o6 AT netnews DOT hinet DOT net> <6e86fs$1kr$1 AT hyperion DOT triode DOT net DOT au>
NNTP-Posting-Host: 203.66.104.168
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Precedence: bulk


telford AT xenon DOT triode DOT net DOT au wrote in
<6e86fs$1kr$1 AT hyperion DOT triode DOT net DOT au>...
>I'm thinking that OpenGL looks like a good idea. Mesa is already free
>and looking for people to write hardware specific support. Worst case,
>the Mesa library can become a repository for graphics know-how that has
>openly available source. This allows newcommers to the industry to get
>up to speed with the concepts quite quickly.
>
> - Tel
>
>

I was originally thinking along the same lines, but I just can't get over
the impression that this is just too slow. GL is cool! GL is great! I know
that GL runs really quickly with acceleration (like on my PII 266MHz 128MB
RAM + SiS Accelerator system at work), but puts me to sleep at home (5x86
(read 486) 133MHz 16MB ram, Tseng ET4000 W32 VGA card and a montiro that's
about to explode).

At home I can run stuff thats made for VGA and get really cool frame rates
and effects. If you people haven't already done so, check out some of the
stuff that the demo scene people are writing. This stuff blows me away
sometimes.

What I really want to do is write stuff that will do that on slow systems,
but take advantage of hardware acceleration if available and add more
content.

If I used MESA, I would end up having to write 2 versions of my code - oen
for openGL and one for vanilla VGA. But on the other hand, if I could just
directly access the trinagle drawing, shading, texture mapping, fog, or
other functions of the accelerator, I could just build a structure with
function pointers, and fill the structure with the appropriate functions
after hardware detection.

For example, lets say I have a function that scans polygons. I already have
routines to do this and draw them out to display memory. But if my program
detected hardware acceleration was available, I could fill inthe structure
with the pointer to a hardware accelerated version of the function.

I have done this already by porting a few of my programs to DirectX, but I
just feel that DirectX is such a hassle that I would rather do things my own
way.


-Chris