Originally Posted by RAMChYLD
Why wouldn't NVidia just support the Wayland protocol?
To my knowledge, on the output side, Wayland only needs the OpenGL or OpenGL/ES protocol with the "render to texture" extension. This is already there in the nvidia linux driver. However, that one only works in the context of a whole X-Server. Many embedded systems come with OpenGL/ES libraries that directly render to the framebuffer (i.e. AMD z430 or the Vivante GPU). However, I heard that nvidia uses a different approach for the Tegra3: it first needs to load and execute some sort of "server" (maybe a minimized X-server? BTW.: starting this server allegedly takes several seconds. With the other GPUs, you can boot into an OpenGL application in half a second - that's pretty important in the automotive area).
For Desktop systems, I don't see an issue Waylind is initially running on top of a full blown OpenGL enabled X-server that provides the required extension. Ok, this contradicts the basic idea of Wayland a little bit, but I personally wouldn't care too much at the beginning. This way, Wayland can get started with nvidia hardware+software without nvidia being involved. And in case Wayland gains momentum over the years, then this may put some pressure on nvidia to provide Framebuffer-OpenGL libraries like the rest of the industry does.