[Octave-bug-tracker] [bug #63995] Octave c++ method .double_value on (intmax ("uint64")-1)

[Hendrik K]

URL:
  <https://savannah.gnu.org/bugs/?63995>

                 Summary: Octave c++ method .double_value on (intmax
("uint64")-1) 
                   Group: GNU Octave
               Submitter: koerhen
               Submitted: Tue 04 Apr 2023 01:10:44 AM UTC
                Category: Interpreter
                Severity: 3 - Normal
                Priority: 5 - Normal
              Item Group: Inaccurate Result
                  Status: None
             Assigned to: None
         Originator Name: 
        Originator Email: 
             Open/Closed: Open
                 Release: dev
         Discussion Lock: Any
        Operating System: GNU/Linux
           Fixed Release: None
         Planned Release: None


    _______________________________________________________

Follow-up Comments:


-------------------------------------------------------
Date: Tue 04 Apr 2023 01:10:44 AM UTC By: Hendrik K <koerhen>
When using the octave c++ method ".double_value" on an octave-value container
with a double representing uintmax-1, the result cannot be represented as the
correct <uint64_t> value, but is always zero.

Example
=======


#include <octave/oct.h>

DEFUN_DLD(ff, args, nargout,
  "-*- texinfo -*-\n\
@deftypefn {} {} ff (@var{V})\n\
@end deftypefn")
{

  octave_value ret;

  octave_stdout << args (0).double_value () << "\n";

  uint64_t t1 = (uint64_t) args (0).double_value ();
  octave_stdout << t1 << "\n";

  uint64_t t2 = static_cast<uint64_t> (args (0).double_value ());
  octave_stdout << t2 << "\n";

  uint64_t t4 = std::trunc(args (0).double_value ());
  octave_stdout << t4 << "\n";

  double t_d = args (0).double_value ();
  uint64_t t5 = (uint64_t) t_d;
  octave_stdout << t5 << "\n";

  double n_d = args (0).double_value ();
  octave_stdout << n_d << "\n";

  uint64_t t6 = static_cast<uint64_t> (n_d);
  octave_stdout << t6 << "\n";

  return ovl (ret);
}

---------------------------------------------------
clear ff;  mkoctfile ("ff.cc");

ff(intmax ("uint64")-1)
1.84467e+19
0
0
0
0
1.84467e+19
0



This was detected as part of the bug https://savannah.gnu.org/bugs/?63992







    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?63995>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/