[Bug c++/49389] New: [C++0x] Wrong value category for pointer-to-member expression with rvalue object expression

["daniel.kruegler at googlemail dot com" <gcc-bugzilla at gcc dot gnu dot org>]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49389

           Summary: [C++0x] Wrong value category for pointer-to-member
                    expression with rvalue object expression
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: daniel.kruegler@googlemail.com
                CC: jason@redhat.com


gcc 4.7.0 20110611 (experimental) in C++0x mode rejects the following program
at the line marked with #:

//----
template<class T>
T&& val();

struct A {};

typedef decltype(val<A>().*val<int A::*>()) type;

template<class>
struct assert_type;

template<>
struct assert_type<int&&> {};

assert_type<type> test; // #
//----

"error: aggregate 'assert_type<int&> test' has incomplete type and cannot be
defined"

Further testing reveals that the deduced type is 'int&' instead of 'int&&'.

According to my reading of 5.5 [expr.mptr.oper] p6:

"The result of a .* expression whose second operand is a pointer to a data
member is of the same value category (3.10) as its first operand."

we have an xvalue expression val<A>() as the first operand, which should result
in an xvalue expression category for the complete pointer-to-member expression.

Referring to 7.1.6.2 [dcl.type.simple] p4, we should fall into bullet 2:

"otherwise, if e is an xvalue, decltype(e) is T&&, where T is the type of e;"

which should have the effect of returning int&&, not int&. Therefore gcc should
accept this program.