How can you learn to invent with spending some time reinventing? In my opinion reinventing the wheel is a often a wonderful task. It helps you understand a hard problem that you would otherwise have serious difficulties with. For example, continuations always baffled me. I ported an existing library from Common Lisp to Clojure. Did I do anything truly new? No. Did I have a much deeper understand of Common Lisp, Clojure and continuations? Yes.

And, sometimes (if you're really lucky) you might find that you have reinvented a better wheel.

I like thinking about how to solve those problems myself, but in this case, I have to write about them. So I thought it would be good to have an idea of the "official" version.

Edit: I suppose I could code something and then put it out for discussion somewhere, before presenting it officially.

I think the best place to learn is cold-hard experience. It doesn't have to be your own experience, of course. Look for open-source projects that utilize similar concepts, and see what they did.

For those questions you can find some of the theory of it in an operating systems design book. As far a best practices you can find those in a few ways experience or working with someone smarter than you.

I find that stackoverflow.com generally has very good answers about stuff like best practices.