Bug 46687

Summary: Class member lookup ambiguity w/ overloaded static members and using declarations
Product: gcc Reporter: Hubert Tong <hstong>
Component: c++Assignee: fabien
Status: ASSIGNED ---    
Severity: normal CC: egallager, fabien, fang, webrown.cpp
Priority: P3 Keywords: accepts-invalid, rejects-valid, wrong-code
Version: 4.5.0   
Target Milestone: ---   
See Also: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=41796
Host: powerpc64-unknown-linux-gnu Target: powerpc64-unknown-linux-gnu
Build: Known to work:
Known to fail: 3.2, 4.1.2, 4.3.2, 4.4.0, 4.5.0 Last reconfirmed: 2012-09-02 00:00:00

Description Hubert Tong 2010-11-27 21:19:15 UTC
In the test case below, there is a set of overloaded static member functions
foo() whose true declarations all come from struct A.

While lookup from C would find foo in both B1 and B2, B1::foo refers to the same
set of entities as B2::foo. There should be no ambiguity; GCC reports otherwise.

The overload candidates for foo should be unambiguously found on the call and
overload resolution should determine that foo(void) is being called.

I also find it odd that the candidates are listed three times.


### Self-contained source (dataa.cpp):
struct A {
   static int foo();
   static int foo(char);
};

struct B1 : A { using A::foo; };
struct B2 : A { using A::foo; };

struct C : B1, B2 { };

enum { X = sizeof C::foo() };


### Command to reproduce:
g++ -c dataa.cpp


### Compiler output:
dataa.cpp:11:19: error: reference to 'foo' is ambiguous
dataa.cpp:3:15: error: candidates are: static int A::foo(char)
dataa.cpp:2:15: error:                 static int A::foo()
dataa.cpp:2:15: error:                 static int A::foo()
dataa.cpp:3:15: error:                 static int A::foo(char)
dataa.cpp:2:15: error:                 static int A::foo()
dataa.cpp:3:15: error:                 static int A::foo(char)


### g++ -v output:
Using built-in specs.
Target: powerpc64-unknown-linux-gnu
Configured with: ../gcc-4.5.0/configure --prefix=/data/gcc --program-suffix=-4.5.0 --disable-libssp --disable-libgcj --enable-version-specific-runtime-libs --with-cpu=default32 --enable-secureplt --with-long-double-128 --enable-shared --enable-__cxa_atexit --enable-threads=posix --enable-languages=c,c++,fortran
Thread model: posix
GNU C++ (GCC) version 4.5.0 (powerpc64-unknown-linux-gnu)
Comment 1 Jonathan Wakely 2010-11-28 01:43:26 UTC
why do you think it's not ambiguous?

C has two bases of type A, so two copies of A::foo()
Comment 2 Jonathan Wakely 2010-11-28 01:57:06 UTC
C has two copies of the name A::foo, as B1::foo and B2::foo.

if C only saw A::foo then it would be unambiguous because the same members would be found, as in this variant:

struct A {
   static int foo();
   static int foo(char);
};

struct B1 : A { };
struct B2 : A { };

struct C : B1, B2 { };

enum { X = sizeof C::foo() };


However, because you have using declarations in B1 and B2 name lookup finds B1::foo and B2::foo ... at least by my reading, which could be wrong
Comment 3 Hubert Tong 2010-11-28 04:23:55 UTC
(In reply to comment #2)
> However, because you have using declarations in B1 and B2 name lookup finds
> B1::foo and B2::foo ... at least by my reading, which could be wrong

It does find B1::foo and B2::foo, but then again, B1::foo and B2::foo refer to
the same functions. In the N3126 wording, the declaration sets for looking up B1::foo and B2::foo are the same. I believe the case presented is valid under both the C++03 and the N3126 wording.

>>>
C++03 subclause 10.2 [class.member.lookup] paragraph 2:
First, every declaration for the name in the class and in each of its base
class sub-objects is considered.
...
Each of these declarations that was introduced by a using-declaration is
considered to be from each sub-object of C that is of the type containing the
declaration designated by the using-declaration. If the resulting set of declarations are not all from sub-objects of the same type, or the set has a nonstatic member and includes members from distinct sub-objects, there is an ambiguity and the program is ill-formed. Otherwise that set is the result of the lookup.
<<<

My understanding is that the resulting set of declarations are all from
subobjects of the same type (the two subobjects, C::B1::A and C::B2::A) and the
set has no nonstatic members (all functions foo() are static member functions).
From this paragraph, we have a set of declarations as the result of lookup:
{ ::A::foo(void), ::A::foo(char) }

Overload resolution then takes place.