Ask YC: Learning Lisp
8 comments
So what's the best way to work through this book?
Get a Lisp and fire up the REPL. Type in the expressions as you read and play with variations.
I also found it helpful to do the exercises.
Is there anything I should keep in mind as I delve into functional programming for the first time?
As you probably know, Lisp isn't a pure functional language. It's its own strange and diverse world. But if you're interested in exploring functional programming in Lisp (a great idea), one important thing is to be aware of when your code has side-effects and get good at eliminating or containing them.
Edit: I'd also recommend staying away from CLOS for a while. Not everyone would agree with that. But my experience has been that focusing on built-in data structures and higher-order functions is the best way to learn to program in the functional style. If you're working through ACL, that's the approach you'll be taking anyway.
Get a Lisp and fire up the REPL. Type in the expressions as you read and play with variations.
I also found it helpful to do the exercises.
Is there anything I should keep in mind as I delve into functional programming for the first time?
As you probably know, Lisp isn't a pure functional language. It's its own strange and diverse world. But if you're interested in exploring functional programming in Lisp (a great idea), one important thing is to be aware of when your code has side-effects and get good at eliminating or containing them.
Edit: I'd also recommend staying away from CLOS for a while. Not everyone would agree with that. But my experience has been that focusing on built-in data structures and higher-order functions is the best way to learn to program in the functional style. If you're working through ACL, that's the approach you'll be taking anyway.
My standard recommendation: begin with the SICP videos, which you can find at http://groups.csail.mit.edu/mac/classes/6.001/abelson-sussma...
I can't think of a more effective way to invest 20 hours.
I can't think of a more effective way to invest 20 hours.
Absolutely!
But although SICP teaches you a great deal about Lisp techniques, it doesn't teach Common Lisp. Abelson and Sussman use Scheme. If you are aware of the differences, it should be no problem, though.
This has been written 10k times here: The video lectures are great, the book is great (It's also free on the internet)! If you don't depend on Common Lisp, this is probably the best Lisp (and programming in genereal) introduction out there. As a follow-up, Paul Graham's "On Lisp" is a great choice.
But although SICP teaches you a great deal about Lisp techniques, it doesn't teach Common Lisp. Abelson and Sussman use Scheme. If you are aware of the differences, it should be no problem, though.
This has been written 10k times here: The video lectures are great, the book is great (It's also free on the internet)! If you don't depend on Common Lisp, this is probably the best Lisp (and programming in genereal) introduction out there. As a follow-up, Paul Graham's "On Lisp" is a great choice.
If you can find a copy of On Lisp (I know it's available online but I just can't read books online).
Yes, the price on Amazon is just insane. I printed the pdf, which worked quite well.
Thanks for the MIT videos link!
Build something cool. Rinse. Repeat.
Doing exercises and plowing through books is great but solving your own problems, to me, is key.
Most books won't even touch the surface of the real things you'll deal with, like how to interface to C or Java for example, or even mention the most important part of developing : other people's code.
My advise FWIW:
First set up sbcl/emacs/slime. This is really not debatable.
Then in no particular order - think holographic distribution:
a. Hang out on #lisp
b. Read comp.lang.lisp
c. Collect and read and use the free (as in liberty) libraries: http://cliki.net http://www.cl-user.net
d. Build software for yourself and your friends
Doing exercises and plowing through books is great but solving your own problems, to me, is key.
Most books won't even touch the surface of the real things you'll deal with, like how to interface to C or Java for example, or even mention the most important part of developing : other people's code.
My advise FWIW:
First set up sbcl/emacs/slime. This is really not debatable.
Then in no particular order - think holographic distribution:
a. Hang out on #lisp
b. Read comp.lang.lisp
c. Collect and read and use the free (as in liberty) libraries: http://cliki.net http://www.cl-user.net
d. Build software for yourself and your friends
I wouldn't recommend hanging out on #lisp for a beginner. In my case it set me in a passive mode and I'd ask a flurry of questions that a little research in the HyperSpec and other appropriate reference documents would have answered.
See, I'm the opposite. I can be quite timid in IRC (though this generally applies to anyone I don't really know), and I always wait until I'm completely stuck to ask a question. Then I feel like a complete tool when someone points me to a link I should have found on my own.
So, I don't think I need to worry about entering passive mode. :)
So, I don't think I need to worry about entering passive mode. :)
| b. Read comp.lang.lisp
For laughing or for crying? There's not much in-between...
For laughing or for crying? There's not much in-between...
False. There's gold in them thar hills!
http://xach.livejournal.com/138588.html
http://xach.livejournal.com/139017.html
http://xach.livejournal.com/138588.html
http://xach.livejournal.com/139017.html
I know you already mentioned and dismissed Practical Common Lisp (free online at http://gigamonkeys.com/book/), but it may be worth a second look once you get a further through ANSI Common Lisp - they're pretty complementary, IMHO.
You may also find the Little Schemer series helpful (though they explicitly cover Scheme, almost everything applies to Lisp) for understanding several aspects of the functional programming mindset. They're also fast reads. Dabbling in languages like OCaml, Haskell, and Erlang will also round out your understanding of functional programming.
While I think SICP is fantastic, it strikes me as more of a "spend years absorbing the deeper ideas in this" book than an "introduction to the tools and idioms of the language" book. Definitely consider it once you're comfortable with Lisp, just probably not first thing.
You may also find the Little Schemer series helpful (though they explicitly cover Scheme, almost everything applies to Lisp) for understanding several aspects of the functional programming mindset. They're also fast reads. Dabbling in languages like OCaml, Haskell, and Erlang will also round out your understanding of functional programming.
While I think SICP is fantastic, it strikes me as more of a "spend years absorbing the deeper ideas in this" book than an "introduction to the tools and idioms of the language" book. Definitely consider it once you're comfortable with Lisp, just probably not first thing.
Check out http://www.cs.northwestern.edu/academics/courses/325/reading... for notes on where Graham's lisp style is notably weird. It also helps to keep in mind that ANSI Common Lisp is not really about ANSI Common Lisp, but an ideal Lisp that doesn't exist. To actually learn Common Lisp (which I can heartily recommend), try Paradigms of AI Programming (despite the title, it's not especially an AI book) or Practical Common Lisp. CLOS is also well-covered by Sonya Keene's CLOS book.
First and foremost it's critically important to be aware of the mistakes in the book. Notable are the following:
You should really use DEFVAR to introduce variables, not SETF as graham does. SETF sets to _places_, and when you use it on undefined variables, the outcome is undefined, but normally a lexical binding is created. I don't know exactly why the book does this.
CLOS is better covered by Sonya Keene's book and Art of the MOP.
Honestly, ANSI CL isn't quite as nice as PCL, but my opinion is that they're both too brief an entry to Lisp, especially for people new to functional programming and Lisp. "A Gentle Introduction to Symbolic Computation" would probably be a better start point for CL, as would those Little Schemer books would be for Scheme.
You should really use DEFVAR to introduce variables, not SETF as graham does. SETF sets to _places_, and when you use it on undefined variables, the outcome is undefined, but normally a lexical binding is created. I don't know exactly why the book does this.
CLOS is better covered by Sonya Keene's book and Art of the MOP.
Honestly, ANSI CL isn't quite as nice as PCL, but my opinion is that they're both too brief an entry to Lisp, especially for people new to functional programming and Lisp. "A Gentle Introduction to Symbolic Computation" would probably be a better start point for CL, as would those Little Schemer books would be for Scheme.
Honestly, ANSI CL isn't quite as nice as PCL, but my opinion is that they're both too brief an entry to Lisp, especially for people new to functional programming and Lisp. "A Gentle Introduction to Symbolic Computation" would probably be a better start point for CL, as would those Little Schemer books would be for Scheme.
I disagree. PCL could have been condensed down to a few presentation slides. It takes a long time to not say much.
I disagree. PCL could have been condensed down to a few presentation slides. It takes a long time to not say much.
PCL could have been condensed down to a few presentation slides
Well, that's a little unfair. It couldn't be condensed to a few slides for people who don't know Lisp. Which is its target audience, after all.
I do agree that PCL isn't as concise as ACL. But we'd know that from the limit theorem on Paul Graham's writing anyway :)
Well, that's a little unfair. It couldn't be condensed to a few slides for people who don't know Lisp. Which is its target audience, after all.
I do agree that PCL isn't as concise as ACL. But we'd know that from the limit theorem on Paul Graham's writing anyway :)
Actually I'd argue that the most important thing is to understand the big ideas, not minutae of "correct" usage.
ANSI CL isn't quite as nice as PCL, but my opinion is that they're both too brief an entry to Lisp
I have to disagree. Those are the two books I learned Lisp from and I didn't find them too brief at all.
I like both books a lot. In some ways they're polar opposites, and I've noticed people tend to prefer the one they're more temperamentally compatible with. ACL has a classical style, precise and formal. PCL is more exploratory and dabbly. If you're the kind of person who wants concepts 1..n to be defined before n+1 is introduced, that's ACL. If you want to see cool examples before grokking all the details, that's PCL. PCL's temperament seems more common than ACL's, as far as I can tell.
I have to disagree. Those are the two books I learned Lisp from and I didn't find them too brief at all.
I like both books a lot. In some ways they're polar opposites, and I've noticed people tend to prefer the one they're more temperamentally compatible with. ACL has a classical style, precise and formal. PCL is more exploratory and dabbly. If you're the kind of person who wants concepts 1..n to be defined before n+1 is introduced, that's ACL. If you want to see cool examples before grokking all the details, that's PCL. PCL's temperament seems more common than ACL's, as far as I can tell.
Wow, thanks for the advice everyone.
Little Schemer is on my to-read list for sure, but I saw ANSI CL and bought it on a whim.
PCL seems recommended by a few people, so maybe I shouldn't have dismissed it so quickly. I opened it up to several pages and saw large blocks of text with very few line breaks, but maybe I just happened to turn to the wrong pages. Perhaps I'll give it another chance.
Again, thanks for the advice everyone!
Little Schemer is on my to-read list for sure, but I saw ANSI CL and bought it on a whim.
PCL seems recommended by a few people, so maybe I shouldn't have dismissed it so quickly. I opened it up to several pages and saw large blocks of text with very few line breaks, but maybe I just happened to turn to the wrong pages. Perhaps I'll give it another chance.
Again, thanks for the advice everyone!
silentbicycle also recommended the Little Schemer books. But for me personally, while they may work for you or someone else... I found that they didn't work for me. Not to say that I have a weight problem, but I love food and food is the theme throughout these books. I like to watch my weight and my sense of control with food is not that stellar; and about three-quarters of the way through the book I realized I wasn't going to make it through to the end or with the sequel: Seasoned Schemer. The food they use to bring their concepts full circle just don't do it for me. Thus, I'm yet another one for SICP or PG's book.
I always recommend Applications Programming in ANSI C by Johnsonbaugh if you're interested in learning C. It's got a great scientific and engineering theme.
While I'm not sure about your area of the world, I know that in my area books at the bricks & mortar stores are ridiculously expensive. Thus, I always do my book buying at places like Amazon and I suggest you do too.. especially if the book you're looking for in the Borders store doesn't carry your desired book. I find the stores like Barnes & Noble to be great places for eye candy... exploring the books they have, sitting in their comfy chairs and making notes on the books I'm going to buy online later. :)
Best of luck to you!
I always recommend Applications Programming in ANSI C by Johnsonbaugh if you're interested in learning C. It's got a great scientific and engineering theme.
While I'm not sure about your area of the world, I know that in my area books at the bricks & mortar stores are ridiculously expensive. Thus, I always do my book buying at places like Amazon and I suggest you do too.. especially if the book you're looking for in the Borders store doesn't carry your desired book. I find the stores like Barnes & Noble to be great places for eye candy... exploring the books they have, sitting in their comfy chairs and making notes on the books I'm going to buy online later. :)
Best of luck to you!
For the beginning this and Practical Common Lisp will serve you well. Lisp isn't that difficult at the beginning, once you go past the parentheses. But soon you'll find yourself looking for an environment in which to do real stuff. Here, if the obvious choices don't suit you, I suggest trying out Clojure. It's a lisp, functional language, and it's 100% embedded in java - meaning it's probably as fast or faster then most lisps, and it has arguably the largest and best organized collection of libraries in existence.
Except the libraries aren't very Lispy. I don't think you can program Clojure without realizing that you are actually programming Java with realistic syntax. (So basically, you need to be good at Lisp and Java. As a result, I wouldn't recommend it for beginners.)
SBCL + SLIME is a nice development environment. CL is far from a perfect programming language, but you can just ignore the stuff you don't like and it comes pretty close.
SBCL + SLIME is a nice development environment. CL is far from a perfect programming language, but you can just ignore the stuff you don't like and it comes pretty close.
So instead I grabbed pg's ANSI Common Lisp (I actually debated between that and Practical Common Lisp, but large pages of text with few line breaks make my brain hurt).
So what's the best way to work through this book? Is there anything I should keep in mind as I delve into functional programming for the first time? I'm not necessarily looking for specific tips on this book, just some pointers on learning lisp in general.