Re: [Scm-discuss] 5f2 regression: r4rstest.scm fails on 32-bit

[Steve VanDevender]

Aubrey Jaffer writes:
 > I compiled scm5 with gcc-5.4.0 -m32 to make a 32-bit executable.  Your
 > examples ran in it without error.  Could it be a GCC optimizer bug?
 > Please try compiling with -O0.

I just tried compiling SCM 5f2 with -O0 on a 32-bit x86 system but still
get the test failure:

(#<CLOSURE float-rw-range-test "r4rstest.scm": () (address@hidden ((success 
#t))  (do ((cnt -323 (+ 1 cnt))) ((> cnt 308) success) (let* ((estr 
(string-append "1.e" (number->string cnt))) (num (string->number estr)) (str 
(number->string num))) (cond ((or (>= (string-length str) 10) (not (equal? 
(string->number str) num))) (set! success #f) (for-each write (list estr num 
str (string->number str))))))))>)  ==> 
;ERROR: "r4rstest.scm": round-quotient: Wrong type in arg1 (0 . 0)
; in expression: (address@hidden>string address@hidden)
; in scope:
;   num
;   estr
;   (cnt . address@hidden)
;   (success . address@hidden)
;   ()  procedure float-rw-range-test
;   (log2 slow-frexp float-precision float-print-test mult-float-print-test 
float-rw-range-test . address@hidden)
;   (f0.0 f0.5 f1.0 f2.0 . address@hidden)
;   ()  procedure test-inexact-printing
; defined by load: "r4rstest.scm"

;STACK TRACE
1; (address@hidden ((estr (address@hidden "1.e" (address@hidden>string 
address@hidden) ...
2; (address@hidden ((cnt -323 (address@hidden 1 address@hidden))) ((#@> 
address@hidden 308) success)  (# ...
3; (address@hidden ((res (address@hidden (address@hidden address@hidden) 
(address@hidden address@hidden address@hidden)  ...
4; (address@hidden ((have-inexacts? (address@hidden (address@hidden>number 
"0.0") (address@hidden ...
...

; program args: ("./scm" "-fr4rstest.scm" 
"-e(test-sc4)(test-cont)(test-delay)(gc)" "-e" "(or (null? errs) (quit 1))")