Why we hate Lisp

MIT handled this very poorly; it's always been very conservative about bulking up departments because things can radically change. Aero/Astro was red hot from, oh, post-WWI to the early 1970s, when it catastrophically crashed without to this date a recovery.

For decades EECS struggled with 40% of the undergraduates and ~1/6th of the institute's resources, in fact when they adopted the old core of 6.001-4 in the early '80s they dropped the "service" course that taught introductory programming for lack of human resources. I don't know how much the faculty was bulked up, the ultimate metric, but before the dot.com crash they committed a quarter billion dollars to a new CS research building that was ugly, very dysfunctional, and expensive to operate (so insecure it requires a full time police officer).

Anyway, I judge that "had their days" differently, at least in the longer term. Most specifically, one of these days a programming error will kill thousands or tens of thousands of people, and at least part of the field will radically change as people get more serious about "quality" software.

(Scare quotes because e.g. Facebook and Boeing have very different, and correct for both of them, definitions of software quality.)

I started with Land of Lisp, but it felt very arcane to me at the time. I would copy in code from the book, and then look back at just jibberish. I gave up on that a few chapters in. Then I made that site in Clojure, one error at a time, pounding like a monkey until it worked. I then read Clojure Programming twice, followed by working through Let Over Lambda. About halfway through Let Over Lambda, I think it "clicked". I finished with an amazing online course in Programming Languages through Brown University. I finished with SICP chapters 1-3.

Nothing worthwhile is ever easy. If you need any help please reach out, you can't waste my time.

1) Install Racket (http://racket-lang.org/) 2) Read Simply Scheme (https://www.cs.berkeley.edu/~bh/ss-toc2.html) by Brian Harvey and Matthew Wright. It's free. Harvey taught the Scheme classes at UC Berkeley for many years, and is an outstanding (and award-winning) teacher (he also has a series of videos that might be helpful).

One of the stated goals of Harvey and Wright's approach is to get the student ready for SICP. It's definitely way above the "add two numbers" level, but is not quite the instant warp factor 10 that you get when you dive into SICP.

If you do this, you should also use the Racket Library (https://www.hashcollision.org/simply-scheme/) that adds all the extra stuff used in the Harvey and Wright text.

Good luck!

The pattern of adding numbers is just an instance of a closed set recursion.

(F e0 ... en) -> (reduce F (e0 ... en) zero-element) -> (F(...(F(F zero-element e0) e1)...) en)

Not just about simple addition, it works with function relating two elements of the same kind to another element of that kind. From 2 arguments to 1, so you can pick another element and relate it to the previous result, so on and so forth until you reduced a list of elements to just one.

Of course it's not an application like crafting a website, or rendering GUI, or manipulating sound samples, but these principles are everywhere.

https://github.com/kanaka/mal

Fully working lisp in 10 (easy to not so easy) steps.

First, you should pick a Lisp-1 because otherwise you'll be hit with "weird" behaviour earlier than necessary.

Next, learn the syntax of your Lisp. Pick one that doesn't have many extensions, as they may become a distraction. Ignore any macros your Lisp supports, leave quasiquote and unquote for later (you can use `quote`, but as a shorthand for the `list` function at first). Learning the syntax won't take long, as it becomes pretty minimal if you omit those things.

Make sure you use an editor that understands your Lisp and has advanced syntax highlighting and auto indent functionality. The difference between a well formatted and highlighted code and a blob of flat text is huge in Lisps, compare this:

    (define (plot-lines data)
    (parameterize ([plot-x-tick-label-anchor 'top-right]
    [plot-x-tick-label-angle 35])
    (let ([out-path (build-path (current-directory) "lines.png")]
    [data (list (tick-grid)
    (discrete-histogram data #:y-min 0 #:y-max 100))])
    (plot data
    #:y-label "sloc"
    #:x-label #f
    #:out-file out-path))))

Similarly, don't ever attempt to balance parens manually, your editor should handle this, along with commands for expanding and shrinking particular expresions (for example, change "(message "%s") something" into "(message "%s" something)" with a single keypress) and other goodies.

After the groundwork is done, start with control-flow constructs (forms in Lisp parlance): constructs for assignement (define/defvar/setq (but not! setf)/etc.), for branching (if/cond/etc.), for grouping (begin/progn/etc.), for looping/iteration (loop/for/dotimes/etc.) and error handling (with-handlers/condition-case/etc.). Learn basic data constructors (list/vector/hashmap/etc.). Experiment with these in the REPL or write some simple scripts using them. Learn how to define functions/procedures and refactor your scripts to use them.

You're basically done at this point. Not in the sense that you know Lisp, but you're ready to learn all there is to Lisp without much hassle. The last obstacle is learning about defmacro and quasiquote/unquote, but that's easy once you're comfortable with lists and iteration. Everything else in most Lisps is built with macros, and at that point you have all the tools needed to write Lisp code, to investigate your Lisp implementation and finally to build your own extensions into the language.

I think that most of the difficulty of learning Lisp stems from learning things in wrong order. Lispers naturally want to show you the most powerful language features to distinguish Lisps from other languages, but they already forgot that these features are built on much simpler things. Or, on the other hand, they drift into lambda calculus and Peano numbers, which smells like a Turing tarpit, where nothing of interest is easy to do.

Anyway, try learning things in that order, maybe it'll help. It worked for me, after years of approaching Lisp and quickly retreating I was finally able to grok it thanks to this approach. YMMV of course.

[EDIT: formatting]