Jon Leech wrote:
> On Tue, Nov 30, 1999 at 11:36:20AM -0800, Bernd Kreimeier wrote:
> > Is there any chance for a separation of enum/data types from
> > function prototypes, header-wise?
> If we were to do this, rather than proliferate headers, I'd prefer
> to stick a #ifndef NEED_GL_PROTOTYPES around the prototypes
Yep, that's actually much less hassle, except shouldn't it be:
#ifndef NO_GL_PROTOTYPES
e.g. the oddball app uses -DNO_GL_PROTOTYPES?
> and let the
> app indicate it doesn't want them. But I'm not sure what it would
> accomplish when the vendor gl.h is fixed on Windows and MacOS - unless
> you're planning to have your own private gl.h on those platforms.
Where needed, yes.
> In fact the enums and prototypes *are* dynamically generated in our
> internal source tree, from ".spec" files describing the enumerants and
> functions and various attributes thereof, such as GLX opcodes and
> dispatch slot offsets. When we update the open source GLX to 1.3
> (shortly), the generator scripts will be included.
OK. I suspect though that they are of limited use, as .spec files don't
get installed, and parsing headers to generate modified headers is
probably different from parsing .spec. I see a chance to create headers
with e.g the NO_GL_PROTOTYPE added using configure/make to locate
GL/gl.h and feed it through a script. Again, not easily on Win32 and
MacOS (GNUwin32 aside), but it would be nice to have for UNIXes.
b.
--
"Problem solving is hunting. It is savage pleasure,
and we are born to it." Thomas Harris, Silence of the Lambs
