I started developing GLAF for my own use. I write tools that must be compatible with several operating systems, including old and new IRIX versions. Old IRIX versions support OpenGL 1.0 only (boosted with many extensions, but 1.0 anyway), while IRIX-6.5.x supports newer OpenGL versions. So, it was necessary to develop some version-independent mechanism for accessing features such as texture objects, copy texture, convolution, color matrix, etc...

Later I realized that this work could be useful for other developers as well, so I decided to give it a name (GLAF, from OpenGL Advanced Functionality), and distribute it as open source.

I can't guarantee that GLAF is a good solution for everybody. Some people may even find it horrendous because it's an extra layer between your application and the real extensions API. I'm not going to defend the validity of GLAF from an advocacy point of view. I'm a GLAF user because I need its services and it suits my needs. Period.

The OpenGL extensions mechanism is good because it lets OpenGL grow very fast, without the need of waiting for newer versions when extra features are developed. It has a drawback, though: the uncertainty of features availability increases the complexity of applications. That's the price you pay for it, and there're several ways of managing it, from manual 'do it yourself' methods to automatized libraries such as GLAF.