warning: assignment to `int' from `double'

warning: assignment to `int' from `double'

[Ralf W. Grosse-Kunstleve]

The following code generates warnings with gcc 3.0.3:

int main()
{
  int i;
  double d;
  i = 0;
  i += d;
  return 0;
}

g++ -g z.cpp -o z

z.cpp: In function `int main()':
z.cpp:6: warning: assignment to `int' from `double'
z.cpp:6: warning: argument to `int' from `double'

Is it possible to suppress this warning without resorting to -w?

Thanks,
        Ralf

Re: warning: assignment to `int' from `double'

[Neil Booth]

Ralf W. Grosse-Kunstleve wrote:-

> The following code generates warnings with gcc 3.0.3:
> 
> int main()
> {
>   int i;
>   double d;
>   i = 0;
>   i += d;
>   return 0;
> }
> 
> g++ -g z.cpp -o z
> 
> z.cpp: In function `int main()':
> z.cpp:6: warning: assignment to `int' from `double'
> z.cpp:6: warning: argument to `int' from `double'
> 
> Is it possible to suppress this warning without resorting to -w?

Have you tried a cast?

Neil.

Re: warning: assignment to `int' from `double'

[Ralf W. Grosse-Kunstleve]

> > operator+= etc. for operations on arrays. The array
> > element types can be custom types with overloaded
> > operators. Any cast buried in the array operator+= would
> > interfere with the semantics of the overloaded custom
> > operators.
>
> I'm not sure why this is relevant; the issue is implicit conversions,
> isn't it?

Consider an idealized array operator+=:

  template <typename ElementType1,
            typename ElementType2>
  std::vector<ElementType1>&
  operator+=(std::vector<ElementType1>& a1,
             const std::vector<ElementType1>& a2) {
    for(std::size_t i=0;i<a1.size();i++) a1[i] += a2[i];
    return a1;
  }

For built-in types this would be fine:

  a[i] += ElementType1(a2[i]);

But for custom types with potentially multiple overloads
for operator+= this construct would result in unexpected
behavior.

> > gcc 3.0.x (incl. 3.0.4) is the only compiler that issues
> > the warning above. (I am also using various EDG derivatives
> > and Codewarrior.)
>
> not a strong argument; you need to reference the standard.

If the standard demands a warning it is counter-productive
in my case. I wish it would be possible to selectively
suppress the warning.

Thanks,
        Ralf

Re: warning: assignment to `int' from `double'

[Ralf W. Grosse-Kunstleve]

Neil Booth wrote:

> Ralf W. Grosse-Kunstleve wrote:-
> 
> > The following code generates warnings with gcc 3.0.3:
> > 
> > int main()
> > {
> >   int i;
> >   double d;
> >   i = 0;
> >   i += d;
> >   return 0;
> > }
> > 
> > g++ -g z.cpp -o z
> > 
> > z.cpp: In function `int main()':
> > z.cpp:6: warning: assignment to `int' from `double'
> > z.cpp:6: warning: argument to `int' from `double'
> > 
> > Is it possible to suppress this warning without resorting to -w?
> 
> Have you tried a cast?

In my application a cast is not an option. I am overloading
operator+= etc. for operations on arrays. The array
element types can be custom types with overloaded
operators. Any cast burried in the array operator+= would
interfer with the semantics of the overloaded custom
operators.

gcc 3.0.x (incl. 3.0.4) is the only compiler that issues
the warning above. (I am also using various EDG derivatives
and Codewarrior.)

Thanks,
        Ralf