Is a compiler allowed to pass a variable by value
when calling a function declared:

void f(const T&);

Clearly this invalidates the use of const_cast<>
inside the body of f (but that evokes undefined
behavior anyway).  Still, this could be an optimization
when T is a basic type.

In fact, it looks like it could only be valid for
a basic type since it would involve a copy ctor
otherwise.  True?


thanks,
marco



[ comp.std.c++ is moderated.  To submit articles, try just posting with ]
[ your news-reader.  If that fails, use mailto:std-c++@ncar.ucar.edu    ]
[              --- Please see the FAQ before posting. ---               ]
[ FAQ: http://reality.sgi.com/austern_mti/std-c++/faq.html              ]

On 27 Oct 1998 17:17:24 GMT, Marco Dalla Gasperina <marcodg@pacifier.com> wrote:

>Is a compiler allowed to pass a variable by value
>when calling a function declared:
>
>void f(const T&);

template <typename T> void f(const T& a, T& b)
{
     cout << a << '\n';
     b++;
     cout << a << '\n';
}

int main()
{
     int i=3;
     f(i,i); // must pass both args by reference!
}


>Clearly this invalidates the use of const_cast<>
>inside the body of f (but that evokes undefined
>behavior anyway).  Still, this could be an optimization
>when T is a basic type.

If the function f(...) is inline, and it is indeed inlined, it
doesn't matter.


>In fact, it looks like it could only be valid for
>a basic type since it would involve a copy ctor
>otherwise.  True?

Basic types must be copied too!  Anyway, as long as the optimization
doesn't change the observable behaviour of a program, it is valid.

--
----------------------------------
Siemel B. Naran (sbnaran@uiuc.edu)
----------------------------------


[ comp.std.c++ is moderated.  To submit articles, try just posting with ]
[ your news-reader.  If that fails, use mailto:std-c++@ncar.ucar.edu    ]
[              --- Please see the FAQ before posting. ---               ]
[ FAQ: http://reality.sgi.com/austern_mti/std-c++/faq.html              ]