f1105031   
   From: daniel.kruegler@googlemail.com   
      
   Am 18.01.2012 21:12, schrieb Michael Smolensky:   
   > Consider an example:   
   >   
   > struct S   
   > {   
   >    S( int ) {}   
   > };   
   >   
   > class C   
   > {   
   > public:   
   >    C() {}   
   >    C( int ) {}   
   >    C( S ) {}   
   >   
   >    int get_int() { return 0; }   
   > };   
   >   
   > int main()   
   > {   
   >    C obj0;   
   >   
   >    int i = 0;   
   >   
   >    C obj1( S( obj0.get_int() ) );  // (1) c'tor C( int )   
   >    C obj2( S( i ) );               // (2) function decl C ()( S )   
   >    C obj3( (S( i )) );             // (3) c'tor C( int )   
   >    C obj4( *&S( i ) );             // (4) c'tor C( int )   
   >   
   >    obj0 = obj1;                    // (5) OK   
   >    obj0 = obj2;                    // (6) error: cannot convert   
   > function ptr to int   
   >    obj0 = obj3;                    // (7) OK   
   >    obj0 = obj4;                    // (8) OK   
   >   
   >    return 0;   
   > }   
   >   
   > Both GCC 4.1 and VC9 treat line (2) as a function declaration, not a   
   > constructor C( S ). Workarounds like (3) and (4) work as expected. Can   
   > someone explain?   
      
   Please read the FAQ for this:   
      
   http://www.parashift.com/c++-faq-lite/ctors.html#faq-10.21   
      
   I wonder that line (4) is accepted, it shouldn't: S( i ) creates an   
   rvalue and the expression &S( i ) would attempt to get the address of   
   it. This should be ill-formed according to 5.3.1 [expr.unary.op] p3,   
   only lvalues (or a qualified-id) are feasible operands of the address   
   operator.   
      
   HTH & Greetings from Bremen,   
      
   Daniel Krügler   
      
      
   --   
         [ See http://www.gotw.ca/resources/clcm.htm for info about ]   
         [ comp.lang.c++.moderated.    First time posters: Do this! ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)