Lambda the Ultimate

SICP was a great eye-opener for me, and it taught me to love Scheme. Common Lisp is all very well, but Scheme was the language that won my heart.

Then I met Haskell. Now, my memories of Scheme are like a third-year frat-boy's memories of his high-school sweetheart^*.

I would suggest that you stick with Scheme at least as far as understanding hygienic macros and continuations. Then move on.

-----

^{* But without the furtive, sticky parts.}

You'll find the text highly regarded by almost all the functional programming communities (Scheme, Haskell, ML, etc..).

That and since you've expressed an interest in learning Haskell, maybe you can help on the SICP translation I'm slowly trying to grind away. Haven't got anyone interested yet (probably won't happen till I get the first 3 chapters get done - and then I'll get mostly criticism), so I'll likely have to cultivate someone who has just finised up Scheme/SICP and wants to move on to ML or Haskell. You're my first attempt at proselytizing someone in that group of candidates. :-)

Seriously, though, you'll find many, if not most, on LtU would recommend the book.

I'd also recommend taking a look at the SICP lectures. Hal Abelson and Gerry Sussman are both extremely good teachers, and the lectures give a you a perspective on the material that's hard to capture on a printed page.

Not a language that I do much with, but there is some activity in the Perl community wrt functional programming. I do think, however, that even if you can manage to do some FP in Perl, it's rather a hard language to learn the paradigm in.

I gave a few moments thought to translating the Little Schemer to Oz, having just done a big chunk of the Reasoned Schemer (the latest in the Little series). But then I figured I'd run into the same kind of translation issues as I would with SICP. I've been working sporadically on SICP for 9 months in Alice (really only a couple days here and there), but it has been secondary to the translation of CTM. Did manage to get chapter 1 translated into Oz and OCaml over the weekend, but it will take some more long weekends before I get much else done.

Anyhow, good luck on your PL journey. LtU suggests more than ample material for reading that can occupy any spare time you happen to make. :-)

[Edit Note: I must've pushed reply to the wrong message, because this was intended as a reply to danmcb.]

It depends on your taste. I think SICP and CTM have more breadth and address more practical issues. SS is more like little polished nuggets of prog. lang. theory. I would probably go SICP or CTM first, then SS to address some of the highly technical issues the first two elide. As for price, I have heard that CTM was once available on the web, and if that were so it is entirely possible that archive.org would have a copy. You didn't hear it here first!

I was initially a little put off by the price of CTM too. But I eventually bit the bullet and bought it. Trust me, it is well worth the money.

[Edit]: Peter, if you happen to be reading this thread: Thank you! CTM is probably one of the most worthwhile textbook purchases I've ever made. You and Seif did a wonderful job putting it together.

Indeed. I've found it well worth the money, and I'm only about 1/3 of the way through it, with an awful lot that I'm really looking forward to reading. Plus it's got a much more clear and readable format. I know a lot of people like the Socratic method used in the Lisper line of books, but I find it almost impossible to read and work with. The Little Schemer is one of the few books that my dog destroyed as a puppy whose passing I did not mourn.

MIT Press judged (correctly, as it turns out) that CTM would be appropriate for the educational market (university courses) and not just for the scientific market (researchers and advanced industry). That is why the price is as low as it is (!). If you've seen other technical books, e.g., by Springer, you know what I mean.

There are also used copies for sale on Amazon and elsewhere (see BookFinder). Some of them are probably student copies. If you check the Wayback Machine you can find the draft we had on the net. The printed version corrects most of the errors in the draft and is more polished.

...to afford CTM, it would be well worth it: I have every confidence that in a year's time, CTM will still be heir apparent (IMHO) to SICP, the content will still be relevant, Mozart-Oz will still be available, and there will still be much for even expert programmers and computer scientists to learn from it.

Thanks for all the responses, I'll try to find an used copy of CTM and save SS for later.

www.bookfinder.com etc.

danmcb: FP is such a different way to work, lots of preconceptions get ripped up in short order. I'm still not sure if or how I will actually use this way of working; I guess that will come out in the wash.

Look at it this way: if all it does is help you be less mystified when you see an anonymous object being created and passed to a method in Java, or be a good judge of when boost::lambda can help you in C++, it will have been worth it. One reason that I heartily recommend CTM is precisely because it isn't about just one paradigm and in fact explicitly rejects purity for purity's sake. The trick to being a competent programmer is having a comprehensive toolbox and, over time, the wisdom to know which tools to use under which circumstances.

This article might be of interest.

I didn't read the thread you linked so hopefully I'm repeating too much, but my impression is that HtDP is very gentle making it a good book for a complete beginner, but rather boring for more advanced programmers. SICP is extremely fun to read, but one of it's "problems" (for some) is the gratuitous mathematical examples/exercises, I would not recommend SICP for a complete beginner as it may send the wrong impression of what programming is and is like. For someone who already knows how to program it is interesting.

...was the poster is a EE, so the engineering bent of SICP actually works in his favour. :-)

shouldn't be a problem, I think. I've quite a bit of "real world" programming experience, what I want is to broaden my mind more than anything else.

I'm pretty used to reading texts and filtering as I go too. With teh amount of stuff there is out there, that's a must ...

"(I started out in that a long time ago, before I realized how much more fun it was to code than design circuits.)"

The funny thing is, when I started studying software in earnest, after quite a lot of years in hardware, I relearned a lot of concepts that were very familiar ...

For instance, when I got into OOP I met the idea of abstracting the API from the implementation. Well, what's one of the first thing you do when you go to design a circuit board? Assign signals to the edge connector. You can then publish that to the rest of the team before you even build anything, if you're reasonably sure of it.

Inheritance as a means of promoting code reuse ... well, when you design a complex system with many boards, you look for common elements that you can build as modules or common circuits or whatever. Once you have them designed and debugged, you reuse them everywhere you can, or use them as a start point for new designs. (Over time you start carrying libraries of such things around with you ...)

I guess those ideas have been used in other industries too, though no exmaples spring straight to mind.

It seems there is no much that's new in the world, anyhow.