Convert everything to 64bit
Posted: Sat Dec 10, 2005 11:11 am
hi,
I would like to change the engine in a way that it uses double values instead of float values. I tried to be clever and thought it would be enough to just modify irrTypes.h by changing
"typedef float f32;" to
"typedef double f32;",
so that all the templates automatically use double values
There were a few problems with the OpenGL Driver, because the OpenGL API seems to need float in some cases, but I think I solved those problems by casting f32 to float were the compiler complained. In the end, it compiled without errors, but when I use the new Irrlicht.dll in the example programs, some of them just crash. HelloWorld works, but does not display the sydney model. The 2DGraphics demo works without problems, and the Meshviewer demo works in software, but looks crappy in OpenGL.
Was it naive to believe it would be so easy, or did I just miss something?
I would like to change the engine in a way that it uses double values instead of float values. I tried to be clever and thought it would be enough to just modify irrTypes.h by changing
"typedef float f32;" to
"typedef double f32;",
so that all the templates automatically use double values
There were a few problems with the OpenGL Driver, because the OpenGL API seems to need float in some cases, but I think I solved those problems by casting f32 to float were the compiler complained. In the end, it compiled without errors, but when I use the new Irrlicht.dll in the example programs, some of them just crash. HelloWorld works, but does not display the sydney model. The 2DGraphics demo works without problems, and the Meshviewer demo works in software, but looks crappy in OpenGL.
Was it naive to believe it would be so easy, or did I just miss something?