Most of those shaders written back then are pretty much worthless now that Renderman has introduced RIS mode (instead of REYES mode), which was introduced last year. Pixar recommends writing shaders in C++ now instead of in RSL, as C++ shaders are just way more performant right now. OSL is pretty mature, I know that both Sony Pictures Imageworks and Double Negative use it in production. Now that Renderman also support OSL I'm pretty sure we're going to see OSL examples posted online. OSL shaders are awesome compared to RSL shaders, as OSL gives all control to the renderer, which allows it to do a lot of optimization (such as using the renderers own importance sampling strategies).
Thanks, I'll have to look even more into OSL (vs. GSL or RSL). I would hardly use the word 'worthless' just because RIS mode was introduced last year. The basics of what the prodecurally generated textures book taught me back then, still apply. Geometry, math, building up patterns in software, etc... Not to mention, the basis of such shaders just need to be translated just as GSL can be relatively simple to translate into OSL as the videos by Thomas Dinges show: https://www.youtube.com/watch?v=4LQXjIDWtz0
