[Bug c++/63458] New: myclassobj >> charac compiles without warning, EVEN though ONLY operator>>(std::string&) exists

[Bug c++/63458] New: myclassobj >> charac compiles without warning, EVEN though ONLY operator>>(std::string&) exists

[mqi30097 at qoika dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63458

            Bug ID: 63458
           Summary: myclassobj >> charac compiles without warning, EVEN
                    though ONLY operator>>(std::string&) exists
           Product: gcc
           Version: 4.9.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: mqi30097 at qoika dot com

Why no warning whatsoever??

//================= go.cpp ======================================
#include <iostream>

class Mystream
{
public:
  Mystream(std::istream& is1) : is{is1} {}
  Mystream& operator>>(std::string&);
  operator bool();
private:
  std::istream& is;
};

Mystream& Mystream::operator>>(std::string& s)
{
  is >> s;
  return *this;
}

Mystream::operator bool()
{
  return is.good();
}

int main()
{
  Mystream ms(std::cin);

  char ch;
  while (ms >> ch) {             // WHY NO WARNING?
    std::cout << ch << '\n';
  }
  return 0;
}

//================= end of go.cpp ======================================

Compile with:
c++ -std=c++11 -o go go.cpp

To "fix" the program:
Change declaration in main from 
  char ch;
to 
  std::string ch; 

Why no warning?
Thanks.

[Bug c++/63458] myclassobj >> charac compiles without warning, EVEN though ONLY operator>>(std::string&) exists

[glisse at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63458

--- Comment #1 from Marc Glisse <glisse at gcc dot gnu.org> ---
You forgot to mark your operator bool as explicit.

[Bug c++/63458] myclassobj >> charac compiles without warning, EVEN though ONLY operator>>(std::string&) exists

[glisse at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63458

--- Comment #2 from Marc Glisse <glisse at gcc dot gnu.org> ---
Actually, I do get a warning:
s.c:29:13: warning: 'ch' is used uninitialized in this function
[-Wuninitialized]

[Bug c++/63458] myclassobj >> charac compiles without warning, EVEN though ONLY operator>>(std::string&) exists

[glisse at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63458

--- Comment #3 from Marc Glisse <glisse at gcc dot gnu.org> ---
Though we could probably add a warning for:

int f(bool b)
{
  return b>>3;
}

since it is unusual to use >> on a boolean, and it would catch your testcase.

[Bug c++/63458] myclassobj >> charac compiles without warning, EVEN though ONLY operator>>(std::string&) exists

[mpolacek at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63458

Marek Polacek <mpolacek at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mpolacek at gcc dot gnu.org

--- Comment #4 from Marek Polacek <mpolacek at gcc dot gnu.org> ---
Yeah, shifting booleans is a weird thing to do.  So is multiplication or
dividing booleans - and we don't warn on that either.