Common LISP: The Language [Paperback]

However, as a previous reviewer pointed out, CLTL is strictly a reference, not a text. If you attempt to use it as an introductory text, you will hate both the book and the language, which will be your loss. To learn the language, I would recommend either "Lisp", by Winston and Horn, or "ANSI Common Lisp", by Paul Graham. After perhaps several years of serious Lisp programming, you will most likely find yourself studying the pages of CLTL, at which point you will appreciate what Guy Steele has succeeded in accomplishing in this slender volume of 1029 pages. Common Lisp is an enormous language, with over 800 built-in functions, many of which have complicated semantics and dozens of keywords that alter those semantics. Considering the daunting task of documenting this language, Steele deserves a medal. (In fact, the book has received various awards.)

Common Lisp was an integral part of several classes that I taught at Caltech for many years; I had students write compilers, interpreters, theorem provers, symbolic manipulators, numerical solvers, graph algorithms, etc. When you attack such a wide range of problems with a single language, you appreciate how rich Common Lisp is, and how well suited it is to all these tasks (yes, even numerical computation). But to get the most out of the language, it's necessary to tap into its more esoteric functions, which is where Steele's book is very handy.

I can think of few topics in the field of computer science that have as rich a history as the language Lisp. It's difficult to present a meaningful view of the language, especially in it's "Common" incarnation, without delving into some of that history. Steele does this exceedingly well in CLTL, although I can understand how it can be off-putting to some; it adds bulk to an already formidable tome, and at times seems to clutter up what ought to be a cut-and-dried presentation of syntax and semantics. However, unless you subscribe to the mystical view that Lisp was created by divine fiat (a theory that is gaining popularity), then you will inevitably have questions as to why things were done in one way and not another. The answers provide insight into language design (or at least the workings of the X3J13 committee), and at times a better mastery of Common Lisp. For those who do not care for such details, Steele sets the digressions off from the main body of the text, making them easy to skip. But I, for one, am happy that this information is recorded somewhere. (If nothing else, it keeps the creationists at bay.)

Like the mathematician Gilbert Strang, who manages to inject humor into the driest of mathematical journals, Steele has found ample opportunities to sneak bits of wordplay and irreverence into CLTL for comic relief. Not only does Steele enliven his program fragments with snippets of pop culture, as in

"(loop for turtle in teenage-mutant-ninja-turtles do..."

but all such references are assiduously listed in the index, which makes it a real hoot to glance through. Listed there are "Mozart, Wolfgang Amadeus", and "Michelangelo (artist)" as well as "Michelangelo (turtle)". We also find things like "goody two-shoes", "oranges, comparing apples with", "square peg in round hole", and numerous foods, including garbanzo beans, ice cream, orange flavor beef, pizza, and peppermint. Under "pasta" we find "see also macaroni". But my favorite index entry is "kludges", which directs us to pages 1 through 971; which is, of course, the entire body of the book, excluding index and appendices. Steele obviously decided to have a little fun, which is understandable considering how dry such books tend to be.

But, before you click this book into your shopping cart, you should realize that the complete text is available on-line, and for free. I'm not sure how Steele swung this with the publisher, but it's out there in the public domain. Finally, I should point out that there are a number of excellent free Common Lisp interpreters available for many different platforms. The best I have found is CLISP, which is maintained primarily by Bruno Haible through the GNU Project. It's reasonably complete and robust.

Happy hacking. May cons be with you.

CLtL is one of the poorest examples of technical communication I've ever seen. Imagine being handed a 1000+ page Microsoft Word document, with the "Track Changes" option turned on. That's what it's like to read this book.

A lot of the prose rambles, with remarkably little content. Most of the time, the text sputters along, the flow will just get started, when it is interrupted by a change bar: "In 1988, ANSI voted to blah blah blah..." Reading this book is like riding the bumper cars at an amusement park. It is both visually and mentally jarring. And the examples? There are plenty of hackneyed examples, but almost no useful ones. Last night I ran across an example where the author flat out admits, in the text, that it is a useless and trivial exercise.

In summary, this book was written by Lisp experts, *for* Lisp experts. Newbies need not apply. This makes it even more remarkable as a bad example, in that it took Steele and others over 1000 pages to tell people something they already know. The bottom line? Aside from AMOP and Keene's excellent book on CLOS, there are *no* decent books for learning Common Lisp, the language. I've read almost all of them, including Graham's.

Make no mistake: Common Lisp is a great language, perhaps *the* most powerful language ever invented. But you sure wouldn't know that from reading this book.

Don't let that stop you from buying this book, though; it is an excellent read, contains a wealth of useful information, and is funny as well. Just make sure you use the standard (available freely on the web as the "Hyperspec") as your authorative reference.