My bile in another post made me take a closer look at the current GL spec for what happens for floating pointer depth buffers.
My next begging, is actually already in NV_depth_buffer_float and it would be nice to see in the spec, and I’d imagine it is not a big deal either. The basic idea is to allow for depth values to be outside of the range [0,1] when the depth buffer is a floating point buffer. Currently glDepthRange’s prototype is
glDepthRange(clampd n, clampd f)
and the GL3.3 spec explicitly says the values are clamped to [0,1]. So, I suggest to get more of NV_depth_buffer_float into the specification to allow for unclamped z-values for floating point buffers by providing a few more functions (just as the NV extensions does).