[Bug c++/11518] New: (b==a && c==a && c!=b) == true

[Bug c++/11518] New: (b==a && c==a && c!=b) == true

[gsinai at yudit dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11518

           Summary: (b==a && c==a && c!=b) == true
           Product: gcc
           Version: 2.96 (redhat)
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: gsinai at yudit dot org
                CC: gcc-bugs at gcc dot gnu dot org,gsinai at yudit dot org

(b==a && c==a && c!=b) 
Tths equation is somtimes true with redhat built gcc 2.6 on Pentiums.
a: double type
b: function call returning a double
c: method call returning a double

/*!
 * \file double.cpp
 * \brief If a==b and b==c then c==b? Not always. Not everywhere.
 * \author Gaspar Sinai <gsinai@yudit.org>
 * \date   2003-07-13.
 * Results:
 *--------------------------------------------------------------
 * Surprises
 *--------------------------------------------------------------
 * Pentium g++ 2.96:
 * setting: a_double: 0.1
 * setting: get_double(): 0.1
 * setting: a_Double.get(): 0.1
 * comparing: a_double == get_double()
 * comparing: a_double == a_Double.get()
 * comparing: a_Double.get() != get_double()
 *--------------------------------------------------------------
 * Pentium g++ 3.2.1:
 * setting: a_double: 0.1
 * setting: get_double(): 0.1
 * setting: a_Double.get(): 0.1
 * comparing: a_double != get_double()
 * comparing: a_double == a_Double.get()
 * comparing: a_Double.get() != get_double()
 *--------------------------------------------------------------
 * Expected
 *--------------------------------------------------------------
 * Alpha g++ 2.96:
 * setting: a_double: 0.1
 * setting: get_double(): 0.1
 * setting: a_Double.get(): 0.1
 * comparing: a_double == get_double()
 * comparing: a_double == a_Double.get()
 * comparing: a_Double.get() == get_double()
 *--------------------------------------------------------------

 * The interesting function:
 * static double
 * get_double(void)
 * {
 *   int i=1;
 *   // If you change i to 1 it will work everywhere. 
 *   return (double (i) / 10.0);
 * }
 */
#include  <iostream>

static double get_double(void);

class Double
{
public:
 double get(void);
 void set(double d);
protected:
 double m_d;
};

double
Double::get (void)
{
  return m_d;
}

void
Double::set (double d)
{
  m_d = d;
}

static double
get_double(void)
{
  int i=1;
  // If you change i to 1 it will work everywhere. 
  return (double (i) / 10.0);
}

int
main()
{
  double a_double = get_double();
  Double a_Double;
  a_Double.set (get_double());

  std::cout << "setting: a_double: " << a_double << std::endl;
  std::cout << "setting: get_double(): " << get_double() << std::endl;
  std::cout << "setting: a_Double.get(): " << a_Double.get() << std::endl;

  if (a_double == get_double())
  {
    std::cout << "comparing: a_double == get_double()" << std::endl;
  }
  if (a_double != get_double())
  {
    std::cout << "comparing: a_double != get_double()" << std::endl;
  }
  if (a_double == a_Double.get())
  {
    std::cout << "comparing: a_double == a_Double.get()" << std::endl;
  }
  if (a_double != a_Double.get())
  {
    std::cout << "comparing: a_double != a_Double.get()" << std::endl;
  }
  if (a_Double.get() == get_double())
  {
    std::cout << "comparing: a_Double.get() == get_double()" << std::endl;
  }
  if (a_Double.get() != get_double())
  {
    std::cout << "comparing: a_Double.get() != get_double()" << std::endl;
  }
  return (0);
}

[Bug c++/11518] (b==a && c==a && c!=b) == true

[pinskia at physics dot uc dot edu]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11518


pinskia at physics dot uc dot edu changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |DUPLICATE


------- Additional Comments From pinskia at physics dot uc dot edu  2003-07-14 12:45 -------
This is known as the excess precision problem, read the references in bug 323.
Closing as a dup of 323.

*** This bug has been marked as a duplicate of 323 ***