> The Art of Computer Programming, by Knuth

Too dense and theoretical, even for someone who enjoys it. Introduction to Algorithms by Cormen, Leiserston, Rivest and Stein is more understandable and practical. It does cover everything useful and is thorough, it just isn't pedantic. [1] http://tinyurl.com/yr257c

> Compilers: Principles, Techniques, and Tools by Aho, Sethi, and Ullman (the dragon book)

Maybe. You might want to try Engineering a Compiler by Cooper and Torczon instead. Again, it's more practical/useful/interesting. But if you're dying to write a parser generator, go ahead and read the Dragon Book. http://tinyurl.com/ys3ql7

> Structure & Interpretation of Computer Programs, by Abelson and Sussman

Available for free: http://mitpress.mit.edu/sicp/full-text/book/book.html

Dunno about the databases book, but the rest are also good.

1. Unless you really, really need those tape sorting algorithms.

>> The Art of Computer Programming, by Knuth > Too dense and theoretical, even for someone who enjoys it. Introduction to Algorithms by Cormen, Leiserston, Rivest and Stein is more understandable and practical. It does cover everything useful and is thorough, it just isn't pedantic.

I completely agree. I own three volumnes of TACP and have very rarely used them. They look good and give instant programmer cred when you utter their titles, but they just haven't proven that useful to me. CLRS, however, is a staple on my programming bookshelf and is well worn.

I'll add Programming Pearls and The Mythical Man-Month to the list that is already growing. Both of those books have given me really good insights over the years.

> I own three volumnes of TACP and have very rarely used them. They look good and give instant programmer cred when you utter their titles, but they just haven't proven that useful to me.

Much of TAoCP is simply worth reading straight-through. I don't use it as a reference all that often either, but when I do, it has never let me down.

I find it sort of weird to talk about these books as if they were O'Reilly titles or something; they're not meant to be "used" so much as studied (slowly!).

> 1. Unless you really, really need those tape sorting algorithms.

Hey, you never know. Tapes have only been truly obsolete for a few years now, for sufficiently large amounts of data. If you're Walmart's DBA and already have millions invested in hardware, those algorithms could come in handy :-).