Just a query on how everyone else does their matrix operations in OpenGL. There doesn't seem to be one _correct_ way of doing it, so I'm keen to see how people go about it.
I have a small engine that I use mainly for 2D stuff. When the shaders are created, the projection matrix is loaded into them. Whenever renderables are drawn, model and view matrices are uploaded to the shader and then the matrix multiplications are performed shader-side. Any local matrix operations - translations/rotations/scaling of the model matrix, for example - are done prior to passing the matrix to the shader.
Seems to work well enough, though there'll no doubt be a more efficient way. For example, I send the model and view matrices to the shader each frame regardless of whether they've been updated or not. I could use a flag here to indicate whether or not they've been updated I guess.
Just wondering how everyone else goes about it. Will move to lounge if necessary.