I had thought about it and come up with something semi-solid (although maybe a little half-baked), but… it doesn’t matter because (and thanks for the patient advice guys) after going through the examples in the Orange Book I have to admit I’m pretty much sold! 
(well, either “sold” or “resigned”…) I understand what Zengar is saying much better now, and really I think that book is all I was looking for, the section on reimplementing fixed functionality is invaluable, at least for a beginner like me.
As far as ShaderGen goes, I did not make any progress in figuring out the difference between [0] and [ i], and it totally boggles my mind. I’m not going to think about it unless I run into the problem in a real application again.
Xmas: I have gone through my application and figured out what states and features I use, and while it is unfortunately not “actually quite low” ( 
), it’s at least a constant, finite list of feature combinations that I know I will be able to reimplement easily, which is good enough for me.
Lord crc: That PDF is really interesting. I’ll read through it after I get my head around the simple stuff. The section on code generation and how they organized all their shaders does help, I’m going to be looking for good ways to writing the same GLSL code over and over again, but keeping it optimized, so it will be handy. Thanks.
For anybody else that reads through this thread and finds themselves agreeing with me, go read the book – I still don’t like having to reimplement all of this stuff from scratch just for tiny shaders, I really feel that it’s an unreasonable amount of hassle, but at least the information in that book takes the edge off.
Thanks,
Jason