Scoping bug in Racket v5.1.1? - racket

Somebody please tell me this is a bug in Racket (v5.1.1),
It seams to be a scoping problem.
(see code and output below)
The return from example-1 shows that x is not
getting incremented as it should and as it does
in example-2 and example-3.
The function example-2 is just a copy of example-1
with the reverse instruction removed from doit.
The function example-3 (as you can see) is a copy
as well but has an additional parameter which by itself
does not change anything but when I test it's value in
the cond statement it shows x as being incremented.
In addition if I don't return anything from example-1
but rather print what it now returns it shows x as
being incremented.
(define (example-1 lst)
(letrec([x 0]
[doit (lambda ()
(reverse
(foldl
(lambda (v store)
(set! x (add1 x))
(cons v store))
'()
lst)))])
(let*([results (doit)])
(list x results)
)))
(define (example-2 lst)
(letrec([x 0]
[doit (lambda ()
(foldl
(lambda (v store)
(set! x (add1 x))
(cons v store))
'()
lst))])
(let*([results (doit)])
(list x results))))
(define (example-3 lst id)
(letrec([x 0]
[doit (lambda ()
(reverse
(foldl
(lambda (v store)
(set! x (add1 x))
(cons v store))
'()
lst)))])
(let*([results (doit)])
(cond [(= 1 id) 'junk])
(list x results)
)))
(printf "example-1 : ~a~n" (example-1 '(a b c)))
(printf "example-2 : ~a~n" (example-2 '(a b c)))
(printf "example-3 : ~a~n" (example-3 '(a b c) 1))
output:
example-1 : (0 (a b c))
example-2 : (3 (c b a))
example-3 : (3 (a b c))

It's a bug -- I filed it with a simpler version of your code. (It's unrelated to the jit though.)
Edit: the bug is fixed now.

Related

LISP function to make all possible pairs from a list?

I'm trying to create a LISP function that creates from a list all possible pairs.
Example of what I'm trying to achieve: (a b c d) --> ((a b) (a c) (a d) (b c) (b d) (c d))
Any advice please? I'm not sure how to approach this problem
Here is a simple solution:
(defun make-couples (x l)
"makes a list of couples whose first element is x and the second is each element of l in turn"
(loop for y in l collect (list x y)))
(defun all-pairs (l)
"makes a list of all the possible pairs of elements of list l"
(loop for (x . y) on l nconc (make-couples x y)))
A recursive solution is:
(defun make-couples (x l)
"makes a list of couples whose first element is x and the second is each element of l in turn"
(if (null l)
nil
(cons (cons x (first l)) (make-couples x (rest l)))))
(defun all-pairs (l)
"makes a list of all the possible pairs of elements of list l"
(if (null l)
nil
(nconc (make-couples (first l) (rest l))
(all-pairs (rest l)))))
Here is a version (this is quite closely related to Gwang-Jin Kim's) which has two nice properties:
it is tail recursive;
it walks no list more than once;
it allocates no storage that it does not use (so there are no calls to append and so on);
it uses no destructive operations.
It does this by noticing that there's a stage in the process where you want to say 'prepend a list of pairs of this element with the elements of this list to this other list' and that this can be done without using append or anything like that.
It does return the results in 'reversed' order, which I believe is inevitable given the above constraints.
(defun all-pairs (l)
(all-pairs-loop l '()))
(defun all-pairs-loop (l results)
(if (null (rest l))
results
(all-pairs-loop (rest l)
(prepend-pairs-to (first l) (rest l) results))))
(defun prepend-pairs-to (e them results)
(if (null them)
results
(prepend-pairs-to e (rest them) (cons (list e (first them))
results))))
the simplest tail recursive variant without explicit loops / mapcar could also look like this:
(defun pairs (data)
(labels ((rec (ls a bs res)
(cond
((null ls) (nreverse res))
((null bs) (rec
(cdr ls)
(car ls)
(cdr ls)
res))
(t (rec
ls
a
(cdr bs)
(cons (cons a (car bs)) res))))))
(rec data nil nil nil)))
CL-USER> (pairs (list 1 2 3 4))
;; ((1 . 2) (1 . 3) (1 . 4) (2 . 3) (2 . 4) (3 . 4))
Tail call recursive solution:
(defun pairs (lst &key (acc '()))
(if (null (cdr lst))
(nreverse acc)
(pairs (cdr lst)
:acc (append (nreverse
(mapcar #'(lambda (el)
(list (car lst) el))
(cdr lst)))
acc))))
Both nreverses are there just for aesthetics (for a nicer looking output). They can be left out.
Try it with:
(pairs '(a b c d))
;; => ((A B) (A C) (A D) (B C) (B D) (C D))
General Combinations
(defun pair (el lst)
"Pair el with each element of lst."
(mapcar (lambda (x) (cons el x)) lst))
(defun dedup (lst &key (test #'eql))
"Deduplicate a list of lists by ignoring order
and comparing the elements by test function."
(remove-duplicates lst :test (lambda (x y) (null (set-difference x y :test test)))))
(defun comb (lst &key (k 3) (acc '()) (test #'eql))
"Return all unique k-mer combinations of the elements in lst."
(labels ((%comb (lst &key (k k) (acc '()) (test #'eql) (total lst))
(let ((total (if total total lst)))
(cond ((or (null (cdr lst)) (zerop k)) (nreverse acc))
((= k 1) (mapcar #'list lst))
(t (let* ((el (car lst))
(rst (remove-if (lambda (x) (funcall test x el)) total)))
(dedup (%comb (cdr lst)
:k k
:total total
:test test
:acc (append (pair el (comb rst :k (1- k) :test test))
acc)))))))))
(%comb lst :k k :acc acc :test test :total lst)))
The number of combinations are calculatable with the combinations formula:
(defun fac (n &key (acc 1) (stop 1))
"n!/stop!"
(if (or (= n stop) (zerop n))
acc
(fac (1- n) :acc (* acc n) :stop stop)))
(defun cnr (n r)
"Number of all r-mer combinations given n elements.
nCr with n and r given"
(/ (fac n :stop r) (fac (- n r))))
We can test and count:
(comb '(a b c d) :k 2)
;; => ((A D) (B D) (B A) (C D) (C B) (C A))
(comb '(a b c d e f) :k 3)
;; => ((B A F) (C B A) (C B F) (C A F) (D C A) (D C B)
;; => (D C F) (D B A) (D B F) (D A F) (E D A) (E D B)
;; => (E D C) (E D F) (E C A) (E C B) (E C F) (E B A)
;; => (E B F) (E A F))
(= (length (comb '(a b c d e f) :k 3)) (cnr 6 3)) ;; => T
(= (length (comb '(a b c d e f g h i) :k 6)) (cnr 9 6)) ;; => T

Difference between '(()) and (cons null null)

I am confused about the difference between '(()) and (cons null null) in scheme.
The code below show that b and c are completely the same thing.
(define (dup2 x)
(let ((d '(())))
(set-car! d (car x))
(set-cdr! d (cdr x))
d))
(define a '(1 2))
(define b (dup2 a))
(define c (dup2 a))
(set-car! b 2)
> c ;; --> (2 2)
However, when I used dup instead of dup2:
(define (dup x)
(let ((d (cons null null)))
(set-car! d (car x))
(set-cdr! d (cdr x))
d))
(define a '(1 2))
(define b (dup a))
(define c (dup a))
(set-car! b 2)
> c ;; --> (1 2)
Variable b and c are different. I have done some experiments, but I haven't understand yet.
The value of d in the first implementation is literal data, and is modified with undefined consequences. To highlight what's happening, consider the following code:
(define (incorrect-list-null-and-x x)
(let ((l '(()))) ; a list of the form (() . ())
(set-cdr! l (cons x (cdr l))) ; (cdr l) is (), so (cons x (cdr l)) should be (x . ()) == (x), right?
; and now l should be (() . (x . ())) == (() x), right?
l))
The expected result is that (incorrect-list-null-and-x n) should return a list of the form (() n), and it does the first time, but successive calls are still accessing the same data:
(incorrect-list-null-and-x 1) ;=> (() 1)
(incorrect-list-null-and-x 2) ;=> (() 2 1)
(incorrect-list-null-and-x 3) ;=> (() 3 2 1)
(incorrect-list-null-and-x 4) ;=> (() 4 3 2 1)
The same problem manifests itself a bit differently in your dup2. Every value returned from dup2 is actually the same pair:
(let* ((x (dup2 (cons 1 2)))
(y (dup2 (cons 3 4))))
(display x)
(display y))
outputs:
(3 . 4)(3 . 4)
because the call (dup2 (cons 3 4)) modifies the same structure that was previously returned by (dup2 (cons 1 2)).
Data literals, like '(()), are meant to be read-only, and modifying it using set-car! or set-cdr! has undefined behaviour. For predictable behaviour, use the (cons '() '()) version if you want to use set-car! or set-cdr! on it.
In particular, cons creates a new cons cell, whereas a data literal usually won't.
Still, for the purposes of implementing dup, why are you using set-car! and set-cdr! anyway? Just use cons directly:
(define (dup x)
(cons (car x) (cdr x)))
In your first code snippet you use (d '(())) which ends up binding a literal to d. You then modify the literal which is generally undefined. In your second code snippet you use (d (cons null null)) which binds d to a newly created 'cons cell' which you then modify. There is no problem modifying that.
Note: you've not defined null. Perhaps you meant '()?

Count of atoms on the each level, Scheme

Please, help me with one simple exercise on the Scheme.
Write function, that return count of atoms on the each level in the
list. For example:
(a (b (c (d e (f) k 1 5) e))) –> ((1 1) (2 1) (3 2) (4 5) (5 1))
My Solution:
(define (atom? x)
(and (not (pair? x)) (not (null? x))))
(define (count L)
(cond ((null? L) 0)
((pair? (car L))
(count (cdr L)))
(else
(+ 1 (count (cdr L))))))
(define (fun L level)
(cons
(list level (count L))
(ololo L level)))
(define (ololo L level)
(if (null? L)
'()
(if (atom? (car L))
(ololo (cdr L) level)
(fun (car L) (+ level 1)))))
(fun '(a (b (c (d e (f) k 1 5) e))) 1)
It's work fine, but give not correctly answer for this list:
(a (b (c (d e (f) (k) 1 5) e)))
is:
((1 1) (2 1) (3 2) (4 4) (5 1))
But we assume that 'f' and 'k' on the one level, and answer must be:
((1 1) (2 1) (3 2) (4 4) (5 2))
How should I edit the code to make it work right?
UPD (29.10.12):
My final solution:
(define A '(a (b (c (d e (f) k 1 5) e))))
(define (atom? x)
(and (not (pair? x)) (not (null? x))))
(define (unite L res)
(if (null? L) (reverse res)
(unite (cdr L) (cons (car L) res))))
(define (count-atoms L answ)
(cond ((null? L) answ)
((pair? (car L))
(count-atoms (cdr L) answ))
(else
(count-atoms (cdr L) (+ answ 1)))))
(define (del-atoms L answ)
(cond ((null? L) answ)
((list? (car L))
(begin
(del-atoms (cdr L) (unite (car L) answ))))
(else
(del-atoms (cdr L) answ))))
(define (count L)
(define (countme L level answ)
(if (null? L) (reverse answ)
(countme (del-atoms L '()) (+ level 1) (cons (cons level (cons (count-atoms L 0) '())) answ))))
(countme L 1 '()))
(count A)
What can you say about this?
Do you know what you get if you run this?
(fun '(a (b (c (d e (f) k 1 5) e)) (a (b (c)))) 1)
You get this:
((1 1) (2 1) (3 2) (4 5) (5 1))
The whole extra nested structure that I added on the right has been ignored. Here is why...
Each recursion of your function does two things:
Count all the atoms at the current "level"
Move down the level till you find an s-expression that is a pair (well, not an atom)
Once it finds a nested pair, it calls itself on that. And so on
What happens in oLoLo when fun returns from the first nested pair? Why, it returns! It does not keep going down the list to find another.
Your function will never find more than the first list at any level. And if it did, what would you to do add the count from the first list at that level to the second? You need to think carefully about how you recur completely through a list containing multiple nested lists and about how you could preserve information at each level. There's more than one way to do it, but you haven't hit on any of them yet.
Note that depending on your implementation, the library used here may need to be imported in some other way. It might be painstakingly difficult to find the way it has to be imported and what are the exact names of the functions you want to use. Some would have it as filter and reduce-left instead. require-extension may or may not be Guile-specific, I don't really know.
(require-extension (srfi 1))
(define (count-atoms source-list)
(define (%atom? x) (not (or (pair? x) (null? x))))
(define (%count-atoms source-list level)
(if (not (null? source-list))
(cons (list level (count %atom? source-list))
(%count-atoms (reduce append '()
(filter-map
(lambda (x) (if (%atom? x) '() x))
source-list)) (1+ level))) '()))
(%count-atoms source-list 1))
And, of course, as I mentioned before, it would be best to do this with hash-tables. Doing it with lists may have some didactic effect. But I have a very strong opposition to didactic effects that make you write essentially bad code.

Define-syntax scheme usage

since yesterday I've been trying to program a special case statement for scheme that would do the following:
(define (sort x)
(cond ((and (list? x) x) => (lambda (l)
(sort-list l)))
((and (pair? x) x) => (lambda (p)
(if (> (car p) (cdr p))
(cons (cdr p) (car p))
p)))
(else "here")))
instead of using all the and's and cond's statement, I would have:
(define (sort x)
(scase ((list? x) => (lambda (l)
(sort-list l)))
((pair? x) => (lambda (p)
(if (> (car p) (cdr p))
(cons (cdr p) (car p))
p)))
(else "here")))
What I could do so far, was this:
(define (sort x)
(scase (list? x) (lambda (l)
(sort-list l)))
(scase (pair? x) (lambda (p)
(if (> (car p) (cdr p))
(cons (cdr p) (car p))
p))))
with this code:
(define-syntax scase
(syntax-rules ()
((if condition body ...)
(if condition
(begin
body ...)))))
What I wanted to do now, is just allow the scase statement to have multiple arguments like this:
(scase ((list? (cons 2 1)) 'here)
((list? '(2 1)) 'working))
but I can't seem to figure out how I can do that. Maybe you guys could give me a little help?
Thanks in advance ;)
If this is an exercise in learning how to use syntax-rules, then disregard this answer.
I see a way to simplify your code that you are starting with.
(define (sort x)
(cond ((list? x)
(sort-list x))
((pair? x)
(if (> (car x) (cdr x))
(cons (cdr x) (car x))
x)))
(else "here")))
Since all the (and (list? x) x) => (lambda l ... does is see if x is a list, and then bind l to x, (since #f is not a list, and '() is not false, at least in Racket), you can just skip all that and just use x. You do not need to use => in case, and in this case it doesn't help. => is useful if you want to do an test that returns something useful if successful, or #f otherwise.
Now, if you want to use a macro, then you're going to need to clarify what you want it to do a bit better. I think that case already does what you want. Your existing macro is just if, so I'm not sure how to extend it.
I found the solution for my question, here it goes:
(define-syntax cases
(syntax-rules ()
((_ (e0 e1 e2 ...)) (if e0 (begin e1 e2 ...)))
((_ (e0 e1 e2 ...) c1 c2 ...)
(if e0 (begin e1 e2 ...) (cases c1 c2 ...)))))
Thank you all anyway :)
Here's a solution :
#lang racket
(require mzlib/defmacro)
(define-syntax scase
(syntax-rules (else)
((_ (else body1)) body1)
((_ (condition1 body1) (condition2 body2) ...)
(if condition1
body1
(scase (condition2 body2) ...)))))
(define (sort1 x)
((scase ((list? x) (lambda (l)
(sort l <)))
((pair? x) (lambda (p)
(if (> (car p) (cdr p))
(cons (cdr p) (car p))
p)))
(else (lambda (e) "here")))
x))
It works in DrRacket. I made three changes to your solution. First, i renamed your sort procedure to sort1 since sort is inbuilt in scheme ( I have used it inside sort1). Second, I have changed the sort1 itself so that the input given will be passed to the procedure returned by scase and you will directly get the sorted result. Third, I have modified the scase syntax extension, so that it will accept the else condition.
>(sort1 (list 3 1 2))
'(1 2 3)
> (sort1 (cons 2 1))
'(1 . 2)
> (sort1 'here)
"here"
I suggest you read "The Scheme Programming Language" by Kent Dybvig. There is an entire chapter on syntactic extensions.

Programatically filling in a letrec in Scheme. Macros or eval?

I'm just playing with an NFA for string recognition. I have a macro that creates a function which consumes input and passes on the rest to some other functions. Because there might be loops in my NFA graph, I'm using letrec to put the whole thing together. Here is some code (been testing in PLT-Scheme):
(define-syntax-rule (match chars next accepting)
; a function that consumes a list of chars from a list l.
; on success (if there's more to do) invokes each of next on the remainder of l.
(lambda (l)
(let loop ((c chars) (s l))
(cond
((empty? c)
(cond
((and (empty? s) accepting) #t)
(else
(ormap (lambda (x) (x s)) next))))
((empty? s) #f)
((eq? (car c) (car s))
(loop (cdr c) (cdr s)))
(else #f)))))
; matches (a|b)*ac. e .g. '(a a b b a c)
(define (matches? l)
(letrec
([s4 (match '( ) '() #t)]
[s3 (match '(c) `(,s4) #f)]
[s2 (match '(a) `(,s3) #f)]
[s1 (match '( ) `(,s2 ,s5) #f)]
[s5 (match '( ) `(,s6 ,s7) #f)]
[s6 (match '(a) `(,s8) #f)]
[s7 (match '(b) `(,s8) #f)]
[s8 (match '( ) `(,s1) #f)])
(s1 l)))
(matches? '(a c))
(matches? '(a b b b a c))
(matches? '(z a b b b a c))
Now, what if I had a simple data-structure to represent my NFA, like a list of lists. e.g.
'((s4 () () #t)
(s3 (c) (s4) #f)
...)
My question is: How would I turn that list into the former letrec statement? I'm not too good with Macros and my understanding is that I probably shouldn't be using eval.
If the list is known at compile time (what I mean is, before your program starts running) then you can use a macro. Otherwise you must use eval.
It's ok. This is one of the good uses for eval. :)
I came up with this macro which seems to do the job
(I'm not an expert either):
(define-syntax nfa
(syntax-rules (let-bindings)
; All the let bindings have been expanded
[(nfa start (let-bindings . bindings))
(lambda (l) (letrec bindings (start l)))]
; Otherwise, expand the next binding
[(nfa start (let-bindings . bindings) (s c n a) . rest)
(nfa start (let-bindings (s (match 'c (list . n) a)) . bindings) . rest)]
; Insert the expanded bindings list
[(nfa start states)
(nfa start (let-bindings) . states)]))
; matches (a|b)*ac. e .g. '(a a b b a c)
(define matches?
(nfa s1 ([s4 ( ) () #t]
[s3 (c) (s4) #f]
[s2 (a) (s3) #f]
[s1 ( ) (s2 s5) #f]
[s5 ( ) (s6 s7) #f]
[s6 (a) (s8) #f]
[s7 (b) (s8) #f]
[s8 ( ) (s1) #f])))
The trick is to use intermediate forms to create "subtitution loops",
and reserve identifiers (cf. let-bindings) to distinguish these intermediate forms
from direct usage of the macro.
I think your problem can be seprate into 2 subproblem:
write a macro that consumes a NFA description and generate a NFA automatically,I call this macro make-NFA
apply make-NFA to a list generated programatically,I call this macro apply-macro
the second subproblem is easy:
(define-syntax apply-macro
(syntax-rules ()
((_ macro ls)
(eval
`(macro ,#ls)
(interaction-environment)))))
;(define ls '(1 2 3))
;(apply-macro if ls)=>2
the first question,I have a DFA sample,you can write a NFA by youself:
(define-syntax make-DFA
(syntax-rules (: ->)
((_ init-state (state : result (symbol -> next) ...) ...)
(letrec
((state
(lambda(sigma)
(cond
((null? sigma) result)
(else
(case (car sigma)
((symbol)
(next (cdr sigma)))...
(else false))))))... )
init-state))))
(define DFA1
(make-DFA q1
(q1 : true (#\a -> q2)
(#\b -> q3))
(q2 : false (#\a -> q1)
(#\b -> q4))
(q3 : false (#\a -> q4)
(#\b -> q1))
(q4 : true (#\a -> q3)
(#\b -> q2))))
(DFA1 (string->list "ababa"));=>#f
well,may be define-macro is a better way to implement apply-macro.