https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67708
Bug ID: 67708
Summary: ambiguous overload for user-defined conversion
Product: gcc
Version: 4.8.0
Status: UNCONFIRMED
Severity: enhancement
Priority: P3
Component: c++
Assignee: unassigned at gcc dot gnu.org
Reporter: crazii.he at yahoo dot com
Target Milestone: ---
Host: Win32
Target: Android
command line: g++ ./sample.cxx
sample.cxx:
----------------------------------------
class String
{
public:
String();
};
class Value
{
public:
operator const String&();
operator int() const;
};
class Concat
{
public:
Concat();
Concat& operator+(const String&);
Concat& operator+(const char&);
};
int main()
{
Value v;
Concat concat;
concat = concat + v + 'A';
}
----------------------------------------
output:
./sample.cxx: In function 'int main()':
./sample.cxx:28:18: error: ambiguous overload for 'operator+' (operand types
are 'Concat' and 'Value')
concat = concat + v + 'A';
^
./sample.cxx:28:18: note: candidates are:
./sample.cxx:19:10: note: Concat& Concat::operator+(const String&)
Concat& operator+(const String&);
^
./sample.cxx:20:10: note: Concat& Concat::operator+(const char&)
Concat& operator+(const char&);
^
please IGNORE linking messages if there's any.
-----------------------------------------
according to C++ standard, this behavior is OK.
but IMHO,
the compiler can pick the best match on 'operator+(const String&)'
since it doesn't need extra conversion after user-defined conversion,
yet 'operator+(int)' need convert int to char after user-define conversion.
So I Suggest it as a ENHANCEMENT, if it doesn't conflict with the C++ standard.
