You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I think we should implement some form of freeglut + subset of OpenGL 1.x (with NV specific extensions).
I did consider using MESA but I think it might be better to keep things minimalistic, as the primary use-case will be unit-tests for Xbox emulation. This means an OpenGL implementation should be rather flexible and easy to understand.
We probably also don't care too much about error-checking or 100% correct behaviour of OpenGL either.
I've started prototyping some (insanely ugly) OpenGL here JayFoxRox#3
Most of our existing code is also close to OpenGL already and could easily be hacked to have simple applications running.
The text was updated successfully, but these errors were encountered:
XGUX is a higher-level-than-XGU, low-level graphics API (grouping GPU commands) that come with XGU. It also takes care to hide some GPU internals (example: converting floats to fixed-point and precomputing terms to be compatible with D3D or OpenGL parametrization).
xgu-gl is a very rough GL implementation on top of XGU. It was designed for my Neverball port, but it's also being used in gish and some other projects of mine. It's not very generic (and it's a non-goal to be generic) and many features don't work as expected: dive into the codebase to adapt it for your project. It is a good base / snippet library for porting OpenGL (or even D3D) applications.
I also did look at Mesa (and even took some of its code), and I think it would have been a good choice. However, the direction that I took with xgu-gl is also fine for me and my projects.
I also agree with my initial statements that OpenGL should be usable for unit testing (also in tandem with low-level APIs); Mesa would have made this more difficult.
Most applications also only use a small subset of OpenGL, so implementing it yourself is very simple and allows for performance optimizations (also why xgu-gl isn't very generic).
I'll keep this issue opened, until XGU is officially "released", and until xgu-gl has better documentation (probably once my Neverball port is complete).
I think we should implement some form of freeglut + subset of OpenGL 1.x (with NV specific extensions).
I did consider using MESA but I think it might be better to keep things minimalistic, as the primary use-case will be unit-tests for Xbox emulation. This means an OpenGL implementation should be rather flexible and easy to understand.
We probably also don't care too much about error-checking or 100% correct behaviour of OpenGL either.
I've started prototyping some (insanely ugly) OpenGL here JayFoxRox#3
Most of our existing code is also close to OpenGL already and could easily be hacked to have simple applications running.
The text was updated successfully, but these errors were encountered: