* Read the colour map `First Entry Index` field. If non-zero, error out as the format specification is vague as to what it does and different readers interpret it differently.
* Rename the variable that holds the colour map entry size as the existing name could be misinterpreted as the size of the pointer, not the thing pointed to.
* Handle images reporting themselves as 15 bits per pixel in the same way as Truevision's example code.
There are two types of Geometry Shaders in OpenGL. One is provided by `GL_EXT_geometry_shader4` and `GL_ARB_geometry_shader4` and requires that certain parameters are set with calls to `glProgramParameteri` before the program is linked. The other is provided by OpenGL 3.2 as a core feature and by the GLES extensions `GL_EXT_geometry_shader` and `GL_OES_geometry_shader` and requires these parameters to be set in the GLSL source itself.
The value of `isGeometryShader4Supported` is being used to determine if the `glProgramParameteri` calls should occur, so it should only be `true` when one of the extensions requiring them is present.
There're cases that the occlusion test result has been retrieved
after the query geometry has been changed, it's the result of the
geometry before the change.
The FBX plugin may fail texture lookup even if the texture is located in
the same folder as the model file. This change tries, as a last resort,
to check if the texture filename is available in the same folder as the
model.
If there's a big view change then it might not be appropriate to wait
for '_queryFrameCount' till the next query is issued, because then
geometry might pop into the view after '_queryFrameCount'.
This is especially important for applications not having a constant
frame rate, but just issue a new frame on demand.
I have found that certain image editing programs export DDS images with both the DDPF_FOURCC and DDPF_RGB pixel format flags. The DDS loader will give preference to the DDPF_RGB flag and try to deduce the format from the RGB bit mask values instead of using the more reliable FOURCC flag. This causes some images to not load, even though the FOURCC flag is supported. I've modified the loader so that it checks for the DDPF_FOURCC flag first, then falls back on deducing the format from the RGB bit masks.
