Hi JS, On Wed, Oct 6, 2010 at 6:58 PM, Jean-Sébastien Guay <[email protected]> wrote: > Looks good. To be safe couldn't we > > #define GLAPIENTRY GL_APIENTRY
I tried this but got a error of multiple definitions, but I couldn't work out why, just changing to blanked worked for me fine, and it mirrors how the OSG declares other C function implementations. I would rather stick with what is known to work with the OSG as part of the OSG build system rather than trying to graft how another library declares and compiles things. >> On the glTexImage3D front it looks like mipmap.c has it's own local >> gluTexImage3D implementation for getting the extension. The code >> looks to me like it should work under Windows, on the OSG side we have >> our own extension methods so we might be able to leverage these, but >> this itself is fraught with integration issues as we build things >> around the idea of a ContextID mapping to extension objects for each >> context. It looks to me that the GLU code in Mesa may well not be >> robust in the presence of multiple contexts... If we can't get it >> working right away then just #if'ing out the glTexImage3D codes might >> be the sensible thing to do. We can come back to this later. > > Well I just get "glTexImage3D identifier not found" and "GL_PROXY_TEXTURE_3D > undeclared identifier". So the problem for me is not even at the usage, it's > just that it can't find them at compilation. I did add GL_PROXY_TEXTURE_3D to my include/osg/Texture3D but I don't think I could have checked it in. I've just booted this machine into another version of Linux so will reboot and check it in. > BTW, I fall into this case: > > #else > > /* Only bind to a GL 1.2 implementation: */ > #define gluTexImage3D glTexImage3D > > #endif > > that's why at the call site it's gluTexImage3D but the error message says > glTexImage3D (no 'u')... > > I've tried removing the #ifdef RESOLVE_3D_TEXTURE_SUPPORT above and seeing > if I can get on with the #ifdef part instead of the #else part, but it needs > an #include <windows.h> where wglGetProcAddress() is declared... > > But as a first step perhaps we can #define RESOLVE_3D_TEXTURE_SUPPORT at the > top of mipmap.cpp, and add #include <windows.h>, see attached file. > > As a side note, while testing this out I saw a warning I've never seen > before: > > warning C4229: anachronism used : modifiers on data are ignored > > Pretty funny. I've never been accused of using anachronisms while > programming before. :-) That was on the line where wglGetProcAddress() was > declared, before adding #include <windows.h>. I will have a look at. Time to reboot. Thaks, Robert. _______________________________________________ osg-submissions mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org
