Hello, Pierre-Francois,
On Monday, March 26, 2018, 1:23:28 AM PDT, Ollagnon, Pierre-Francois
<[email protected]> wrote: I have saw a strange behavior
previously: A function was correctly running (processing a big curve) in the
“main_thread” but was ending by a segfault when launch in a “secondary thread”
(ie a thread created by instantiate:pthread) looking at the core file it was
due to recursive call.Best regards,
This behavior could be due to differences in the default stack size given to
processes and threads. On Linux, for example, processes could be allocated a 8
MB stack but threads could be only allocated a 2 MB stack. I can easily imagine
a recursive procedure running to success in 8 MB but failing with only 2 MB.
Best Regards,Joe
From: Joseph Donaldson [mailto:[email protected]]
Sent: Friday, March 23, 2018 10:54 PM
To: Ollagnon, Pierre-Francois <[email protected]>;
[email protected]
Subject: Re: [bigloo] (Probably a bug) Segfault using #t / #f instead of numbers
Hello, Pierre-Francois,
In your second example, you are testing equality between numeric values and the
booleans #t and #f. Neither of these equality checks will every return true
(the only value equal to #t is #t, likewise for #f; see r5rs ), and hence your
function loops until it runs out of stack space and throws a seg-fault. Try the
following:
(define (combination_seg-fault n s lst proc-to-apply)
(cond ((equal? n 0) (proc-to-apply lst)) ;;Final
((equal? s 1) (proc-to-apply (append (make-list n #f) lst))) ;;Final
((equal? n s) (proc-to-apply (append (make-list n #t) lst)))
;;Final
(else (begin
(combination_seg-fault (- n 1) (- s 1) (cons #t lst)
proc-to-apply)
(combination_seg-fault (- n 1) s (cons #f lst)
proc-to-apply)))))
Best Regards,
Joe
|
|
| |
The R5RS standard - The CHICKEN Scheme wiki
|
|
|
On Fri, Mar 23, 2018 at 9:07 AM, Ollagnon, Pierre-Francois
<[email protected]> wrote:
Hello all,
Is there any reason for which I get a different behaviour on this 2 functions:
1:=> (define (combination n s lst proc-to-apply)
(cond ((equal? n 0) (proc-to-apply lst)) ;;Final
((equal? s 1) (proc-to-apply (append (make-list n 0) lst))) ;;Final
((equal? n s) (proc-to-apply (append (make-list n 1) lst)))
;;Final
(else (begin
(combination (- n 1) (- s 1) (cons 1 lst)
proc-to-apply)
(combination (- n 1) s (cons 0 lst) proc-to-apply)))))
1:=> (combination 4 2 '() print)
(0 0 0 1)
(0 0 1 0)
(1 1 0 0)
(1 1 0 0)
1:=> (define (combination_seg-fault n s lst proc-to-apply)
(cond ((equal? n #f) (proc-to-apply lst)) ;;Final
((equal? s #t) (proc-to-apply (append (make-list n #f) lst))) ;;Final
((equal? n s) (proc-to-apply (append (make-list n #t) lst)))
;;Final
(else (begin
(combination_seg-fault (- n 1) (- s 1) (cons #t lst)
proc-to-apply)
(combination_seg-fault (- n 1) s (cons #f lst)
proc-to-apply)))))
= >(combination_seg-fault 4 2 '() print)
*** ERROR:bigloo:
`segmentation violation' exception -- raised
Since it is the same code, just dealing with other values I was expecting to get
(#f #f #f #t)
(#f #f #t #f)
(#t #t #f #f)
(#t #t #f #f)
Note that I get the same behavior in compiled and interpreted mode.
Best regards,
Pierre-Francois
