There's a book, by a smart guy named Sippser, on compubility theory. It's very easy to read for a theory book, in earnest. Try HopCraft and Ulman if you want a harder read.

You know the language, now know more about languages... why they work and what they can do. It's the reason I stopped posting to node questions... it's all been done before. Unless I can prove that a computer cannot do it or do it in some tractable time, it's easy to do.

Writing an OS, a database, a programming language, all easy. Want me to do it? Sure, I can give you /something/ given enough time, money or man power. What cannot be done or has not been done before is the next step, if you want to keep learning. It includes includes a lot of other semi-practical knowledge: compilers, operating systems, computer architecture, language theory, computibility thoery, data structures etc.. It's nothing you can't learn on the fly. The harder stuff, the real theory stuff and inventing stuff, that's a lifetime road.

But I speak with the slant of an achedemic. I wish I thought of mp3 (storing the audible band), or some of the np-complete proofs (tetris is np-complete), or quantum mechanics. All of this Class::DBI, Rec::Descent stuff, valuable and great, but nothing that hasn't been done before.