Quote:

It's a bit dangerous to assign a specific percentage number to this, and I think the 50% number came from the number of functions that are required in OpenGL 2.1, and which also exist in TinyGL.

However, shaders is definitely the big one here. Once that one is done, we'd be pretty close to OpenGL 2.1 functionality, in the sense that the remaining missing functionality will be relatively small in comparison, at least if we disregard the functionality that's very rarely used, if ever, and which I'm not currently planning to implement.

Actually, of the 5 things mentioned, only really shaders has a direct relationship to achieving OpenGL 2.1 functionality. Framebuffer objects aren't part of (core) OpenGL until OpenGL 3.0, and OpenGL programs were never part of (core) OpenGL, it only exists as an optional extension. Implementing the fixed-function pipeline as shaders could count towards OpenGL 2.1 functionality, but it would mainly fix existing functionality to work correctly in all cases, rather than add new functionality. The R600 driver is of course unrelated to reaching the OpenGL 2.1 feature level, as that can be done with the existing R300 driver.

But to attempt to answer your question, and please don't understand this as any sort of exact estimate, but I'd say that shaders probably represent something like 60% of the work to go from MorphOS' previous state to OpenGL 2.1 (again, with the exceptions mentioned above), and implementing the fixed-function pipeline as shaders would probably represent another 20%. The remaining 20% is spread over various not-too-difficult tasks, which range from "pretty simple to do" to "this will take a bit of time, but it's mostly just tedious work, there's no real challenge in it".