David Bergman wrote:
> the problem I had with the NotReallyType stemmed from my Mingw
> distribution (one of several I use...) had a version of GL.h that
> wouldn't compile. [...]
Huh? I'm using MinGW, too, and my <GL/gl.h> seems to work. What is
exactly the problem (config.log, MinGW version, etc.)?
> The moral of this story is that even seemingly unimportant include
> files need to be valid, else strange results can occur.
Well, <GL.gl.h> *is* important, at least if you call configure with
--enable-hopengl. Furthermore, it is correctly guarded by an #ifdef
for systems which don't have such a header at all. Guarding against a
broken system header takes some more work. Could people with such a
broken header please try the following in aclocal.m4:
1) Replace
#ifdef HAVE_GL_GL_H
# include <GL/gl.h>
#endif
with
#ifdef HAVE_GL_GL_H
# ifdef HAVE_WINDOWS_H
# include <windows.h>
# endif
# include <GL/gl.h>
#endif
2) Replace
#ifdef HAVE_GL_GL_H
# include <GL/gl.h>
#endif
with
#ifdef HAVE_GL_GL_H
# include <GL/glu.h>
#endif
> One problem I have right now is that do use OpenJade, which the
> configure script finds, but it cannot find the proper DocBook SGML
> files. Well, well.
Processing DocBook documentation portably seems to be nearly
impossible, I have to fix the corresponding autoconf magic almost
every time a new SuSE Linux is released. Have a lot of fun... :-]
Cheers,
S.
_______________________________________________
Cvs-ghc mailing list
[EMAIL PROTECTED]
http://www.haskell.org/mailman/listinfo/cvs-ghc
