This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Remove algo code duplication


[Jonathan already answered most of it]

On Wed, 2 May 2012, François Dumont wrote:

2. Compatible with existing code, code submitted by Christopher last time will still compile. To do so I used std::iterator_traits<>::reference type so that if the iterator is not const then the operator do not have to take const reference or to be const qualified neither. For the same reason I have also avoided some const qualifiers on the introduced functors.

It probably doesn't matter, but the difference with the current code can be seen. With -fno-elide-constructors you perform copies, operator< is always given an lvalue, etc.
You mean functor copies, right ? I expect the compiler to optimize them away even if there are options to forbid him to do so.

I mean value copies, performed by the functor.


#include <iostream>
struct A {
  A(){std::cout << "A()\n";}
  A(A const&){std::cout << "A(A const&)\n";}
  A(A &&){std::cout << "A(A &&)\n";}
};
bool operator<(A,A){return true;}
struct less {
  bool operator()(A a,A b)const{return a<b;}
};
struct Iter {
  A operator*()const{return A();}
};

int main(){
  Iter x;
  std::cout << "hello\n";
  *x < *x; // all copies elided
  std::cout << "pause\n";
  less()(*x,*x); // 2 copies can't be elided
  std::cout << "bye\n";
}

hello
A()
A()
pause
A()
A()
A(A const&)
A(A const&)
bye


I am not saying it is necessarily a bad thing, just that the difference is noticable. Note that the functor less posted by Jonathan would keep 2 move constructors instead of the 2 copy constructors.


(note: I am not sure that's what I had in mind when I wrote that email, I can't remember exactly)

--
Marc Glisse


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]