Wrong FP errors on Windows 64

Bug #1458160 reported by 3b
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
SBCL
Fix Released
Undecided
Unassigned

Bug Description

on x8664 windows, 1.2.11.70

(macrolet ((fp (x)
             `(format t "~&~a: ~s~%" ',x (nth-value 1 (ignore-errors (eval ',x))))))
  (fp (cosh 90))
  (fp (/ 1d0 0d0))
  (fp (cosh 90))
  (fp (/ 0d0 0d0))
  (fp (cosh 90))
  (fp (/ 1d0 0d0)))

==>

(COSH 90): #<FLOATING-POINT-OVERFLOW {10086D8873}>
(/ 1.0d0 0.0d0): #<DIVISION-BY-ZERO {1008760A73}>
(COSH 90): #<DIVISION-BY-ZERO {10087E90E3}>
(/ 0.0d0 0.0d0): #<FLOATING-POINT-INVALID-OPERATION {10088713E3}>
(COSH 90): #<FLOATING-POINT-INVALID-OPERATION {10088F99F3}>
(/ 1.0d0 0.0d0): #<FLOATING-POINT-INVALID-OPERATION {1008981D23}>

expected OVERFLOW for all calls to COSH, DIVISION-BY-ZERO for (/ 1d0 0d0)

causes
 "Failure: compiler.pure.lisp / NO-OVERFLOW-DURING-ALLOCATION"
in tests due to test expecting OVERFLOW but getting INVALID due to earlier tests.

Revision history for this message
Stas Boukarev (stassats) wrote :

No longer happens.

Changed in sbcl:
status: New → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.