On Thu, 9 Sep 1999, Thomas Roell wrote:
> > Also, an emerging use of wglGetProcAddress() is to return pointers
> > to OpenGL 1.2 core functions. While this exact scenario is unlikely to
> > arise on Linux, we should still be very clear up front about whether
> > non-EXT functions can be queried.
>
> Please let's avoid this. It would be intresting to know as to why
> people are using this approach.
<sigh> I seem to have spent all day patiently explaining this.
A Story.
~~~~~~~~
One bright summer morning, the ARB pronounces the existance of
OpenGL 1.3. It adds "glRealisticShadowsForFree()".
Two days later the OpenSource guys have implemented it on Mesa
and we have a usable OpenGL 1.3.
* 50% of Linux users read about it on Slashdot download it
immediately - flooding the www.mesa3D.org server.
* 1% of them are running XiG OpenGL 1.2 and they get an OpenGL 1.3
upgrade (let's be generous) a month later.
* 49% of Linux users install RedHat 8.9 six months later and
*STILL* have OpenGL 1.2.
OK, I'm writing a new game. I'd like to use glRealisticShadowsForFree.
What do I do?
If glGetFuncAddress is defined to return valid results for gl*
calls, I can say:
glGetString ( GL_VERSION ) ;
...parse string...
if ( we_are_gl_1_point_3 )
shadows = glGetFuncAddress ( "glRealisticShadowsForFree" ) ;
else
shadows = NULL ; /* Doesn't exist in OpenGL 1.2 :-( */
...later...
if ( shadows != NULL )
(*shadows)() ;
...if glGetFuncAddress is defined NOT to return gl* (non-extension)
addresses (for some unfathomable reason, 80% of the people on this
list have been asking for it to work in this crippled way)
THEN, all I can do is this:
#ifdef GL_VERSION_1_3
glRealisticShadowsForFree() ;
#else
/* Tough luck */
#endif
...which means that I either have to hold off shipping my game
until OpenGL 1.3 finally makes it onto 90% of the computers in
the world...OR...I have to ship two binaries - one for OpenGL
1.2 and another for 1.3 (and eventually, yet another for 1.4...)
...OR...I could do the disgusting Windoze trick of using dlopen/dlsym.
But, the last option is out because:
"We Are Not At Home To Mr Kludge Today"
OK.
Now, will *SOMEONE* give me ANY REASON why it's bad to have
glGetFuncAddress return the address of a regular gl* function.
So far, I have two possible reasons:
* This morning someone said the lookup table would be too big.
I computed it - it's ~5kbytes longer if you do it the
naive { "glBegin", glBegin } way, ~2kbytes if you go
for a clever hashtable approach so glGetFuncAddress is
**FAST**.
Either way makes libMesaGL about a half percent larger
overall.
* It'll take poor Brian too long to type in all those function
names tonight when he implements it.
I'm betting it's the second reason - and I sympathise.
Steve Baker (817)619-2657 (Vox/Vox-Mail)
Raytheon Systems Inc. (817)619-2466 (Fax)
Work: [EMAIL PROTECTED] http://www.hti.com
Home: [EMAIL PROTECTED] http://web2.airmail.net/sjbaker1
