Some more progress

README.md

@@ -1,13 +1,11 @@
-# hello
+# SICP
 
-This repository is where I keep my solutions to the problems in SICP
+This repository is where I keep my solutions to the problems in SICP.
 
 I'm currently working through the book and wanted to keep track of my progress somewhere.
 Note: I won't be solving ***ALL*** problems. I just solve *most*, I will probably ignore
 those that seem like they would take too long (or ones that require several PhD's to solve).
 
-Note: these are all racket files, I am also learning racket at the same time (and using
+Note that some of the solutions were done in racket. I started with racket, but eventually
-DrRacket) so beware of that. I will mostly try to stick to racket, but there may be 
+I switched to common lisp. Right now I'm using Emacs with SLIME, and will hopefully be
-times where I switch to `#lang r5rs` or `#lang sicp` for compatability with the book.
+continuing with that for the rest of the book.
-
-Also, I will put related-seeming or close-together problems in the same file.

sec-2-4-3-data-directed.lisp

@@ -183,6 +183,7 @@
 ;; to the table as well.
 
 ;; Note: this problem is sometimes referred to as the "expression problem."
+;; especially within compiler development.
 
 
 

sec-2-5.lisp

@@ -0,0 +1,60 @@
+;; 2.77
+;; obviously magnitude isn't defined yet on the complex type,
+;; it is only defined in the internal types, i.e. the secondary types.
+;; this means that the first apply-generic will immediately fail.
+;; the reason Alyssa's fix works, is because this makes the
+;; maginute function dispatch to itself when seing the type
+;; '(complex). Then, magnitude will be called with the
+;; internal type - e.g. '(rectangular 3 4), which gives
+;; the correct answer.
+
+;; 2.78
+
+(defun type-tag (datum)
+  (cond
+    ((consp datum) (car datum))
+    ((numberp datum) 'number)
+    (t (error "unkown type"))))
+
+(defun contents (datum)
+  (cond
+    ((consp datum) (cdr datum))
+    ((numberp datum) datum)
+    (t (error "unknown type"))))
+
+;; assuming the dispatch function actually calls the function with
+;; the contents of each datum, then you should now be able to just
+;; insert #'+, #'*, #'- and #'/
+
+;; 2.79
+
+;; 2.81
+;; First, apply-generic is called.
+;; 'exp is not defined on complex numbers. Therefore, the apply-generic
+;; function somewhat mistakenly enters the second branch, and first tries
+;; to find a coercion from the first argument's type to the second argument's
+;; type. It finds this coercion function. But then it calls:
+;; (apply-generic op (t1->t2 a1) a2)
+;; this is a problem, because it now enters an infinite loop. Every time,
+;; apply-generic will try to find the 'exp method, not find it, and try
+;; to coerce the first argument to the second argument's type.
+;; It will succeed, and call itself again.
+;; Since this is a tail call, it will be optimised and this will be reduced
+;; to an infinite loop.
+;;
+;; Something can be done about it. We can simply compare the tags of
+;; the objects before trying to coerce - and we can simply raise an error
+;; if both objects have been coerced to the same type and there is still
+;; no function found. Coincidentally, this would also prevent Louis Reasouner's
+;; previous infinite loop.
+
+
+
+;; In order to be extra spicy, I will be completing the exercises in
+;; the symbolic algebra section through the use of CLOS.
+;; I.e. I will actually build an AST of these objects, and use
+;; methods to parse, interpret and modify them.
+
+;; Tomorrow. I will do that tomorrow.
+
+

sec-3-3.scm

@@ -0,0 +1,134 @@
+
+;; I am not drawing box-and-pointer diagrams lmao
+;; still, to answer 3.12:
+
+;; first:
+;; (cdr x)
+;; => '(b)
+;; because x is not modified in the call to append
+
+;; however, append! modifies the underlying data structure
+;; held in x, hence the second (cdr x) returns:
+;; => '(b c d)
+
+(define (append! x y)
+  (set-cdr! (last-pair x) y)
+  x)
+
+(define (last-pair x)
+  (if (null? (cdr x)) x (last-pair (cdr x))))
+
+;; don't run these twice, you'll create a circular list.
+;; I lost a good many REPL's to this.
+(define x '(a b))
+(define y '(c d))
+(append! x y)
+x ; => '(a b c d)
+(cdr x) ; => '(b c d)
+
+
+;; 3.13:
+;; oh! how very nice - a question about the very thing I just
+;; wrote about.
+
+;; the REPL hangs. It tries to traverse to the end of a list
+;; that doesn't have an end. Like most human lives, it dies
+;; a meaningless death in search of that which does not exist,
+;; a search that goes round and round and round forever.
+
+;; I.e. we constructed a cyclical list.
+
+;; 3.14:
+;; This is a list reversal procedure.
+;; The input list is destructively reversed in-place, i.e. with
+;; no allocations.
+
+;; v will still refer to the same cons cell, so its value is
+;; '(a)
+;; w will refer to the new head of the (now-reversed) list.
+;; '(d c b a)
+
+(define (mystery x)
+  (define (loop x y)
+    (if (null? x)
+	y
+	(let ((temp (cdr x)))
+	  (set-cdr! x y)
+	  (loop temp x))))
+  (loop x '()))
+
+;; this procedure is useful because its sometimes very convenient
+;; to generate data into a list through cons cells, then reverse
+;; it at the end (through an efficient reversal procedure like this)
+
+;; 3.16:
+
+(define (count-pairs x)
+  (if (not (pair? x))
+      0
+      (+ (count-pairs (car x))
+	 (count-pairs (cdr x))
+	 1)))
+
+(count-pairs '(a b c)) ; => 3
+
+(define fourbase (cons 1 '()))
+(define four (cons fourbase (cons fourbase '())))
+(count-pairs four) ; => 4
+
+(define sevenbase (cons 1 '()))
+(define sevenmid (cons sevenbase sevenbase))
+(define seven (cons sevenmid sevenmid))
+(count-pairs seven) ; => 7
+
+;; and for the great finale...
+(define finale-a (cons 'a '()))
+(define finale-b (cons 'b finale-a))
+(define finale-c (cons 'c finale-b))
+(set-cdr! finale-a finale-c)
+;; (you can do this with a single cons cell, actually)
+
+;; 3.17
+;; it seems that eq? really does simply check for
+;; pointer equality - if literally the same cons
+;; cell is passed twice it returns #t but if two
+;; cons cells with the same values are created
+;; they measure #f.
+
+;; it's honestly very convenient to know this
+;; it simplifies and clarifies a lot of things.
+
+;; we can use this to essentially build a list of
+;; all pairs we have ever visited. We can have this
+;; list behave like a set - which we defined before.
+
+;; I used a hash table instead because they were available.
+;; and efficient, I guess.
+
+;; You can count the number of pairs without falling
+;; for cycles by just refusing to traverse a cell
+;; that has already been traversed, though this
+;; requires extra storage.
+;; I'm gonna count this for 3.18, as the logic
+;; is incredibly similar.
+
+
+(define (count-pairs-2 x)
+  (define table (make-eq-hashtable))
+  (define (mark-visited! c)
+    (set-cdr! (hashtable-cell table c #t) #t)
+    #t)
+  (define (visited? x)
+    (eq-hashtable-ref table x #f))
+  (define (loop x)
+    (when (and (pair? x) (not (visited? x)))
+      (mark-visited! x)
+      (loop (car x))
+      (loop (cdr x))))
+  (loop x)
+  (vector-length (hashtable-values table)))
+
+
+
+
+