In GL3 or ES2, you load a matrix uniform into the shader, and transform there. As long as your shader runs in hardware, you still have hardware TCL. It's up to you to produce the matrix you load, though. glRotate etc are no longer available to do the math for you.

In general, the trend for the new API is to remove all redundant functionality, and remove all slow paths (things hardware doesn't natively accelerate). This leaves the GL as essentially a memory manager, a triangle rasterizer, and a shader compiler. Almost everything else is gone.

Unfortunately (IMHO) there is no new official GLU or GLUT to replace the removed functionality with a standardized, optimized implementation of all the convenience utility functions. So everybody gets to roll their own. I think this makes the new API harder for beginners to pick up.

arekkusu Wrote:Unfortunately (IMHO) there is no new official GLU or GLUT to replace the removed functionality with a standardized, optimized implementation of all the convenience utility functions. So everybody gets to roll their own. I think this makes the new API harder for beginners to pick up.

Yeah, that sucks.

At any rate, I understand the desire to cut out all the fluff and old stuff. Shaders are clearly the future, and I'm already finding myself skipping lots and lots of OpenGL calls with ES. Plus I have to say that I appreciate ES's slimmed down API.

Could I have been more wrong? I guess I'll have to read the OpenGL 3.0 spec.
(In my defense, scaling objects based on their z-coordinate is, nevertheless, a "silly" idea.)