From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-help-return-44556-listarch-gcc-help=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 23827 invoked by alias); 17 Feb 2011 14:08:34 -0000
Received: (qmail 23816 invoked by uid 22791); 17 Feb 2011 14:08:33 -0000
X-SWARE-Spam-Status: No, hits=-2.3 required=5.0	tests=AWL,BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,FREEMAIL_FROM,RCVD_IN_DNSWL_LOW,T_TO_NO_BRKTS_FREEMAIL
X-Spam-Check-By: sourceware.org
Received: from mail-iy0-f175.google.com (HELO mail-iy0-f175.google.com) (209.85.210.175)    by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Thu, 17 Feb 2011 14:08:28 +0000
Received: by iyj8 with SMTP id 8so2469976iyj.20        for <gcc-help@gcc.gnu.org>; Thu, 17 Feb 2011 06:08:27 -0800 (PST)
MIME-Version: 1.0
Received: by 10.42.241.70 with SMTP id ld6mr2860309icb.124.1297951706961; Thu, 17 Feb 2011 06:08:26 -0800 (PST)
Received: by 10.42.230.68 with HTTP; Thu, 17 Feb 2011 06:08:26 -0800 (PST)
In-Reply-To: <20110217131616.GU5274@axel>
References: <201102171203.53710.pascal@francq.info>	<AANLkTi=MyLQc-PkZ2CjPmbO9Lv1Oia8Dfvmz-SbNgHXa@mail.gmail.com>	<20110217131616.GU5274@axel>
Date: Thu, 17 Feb 2011 16:30:00 -0000
Message-ID: <AANLkTinRGTtGmm_Vok=1Cnit58G4tb4wPFjFpDTXfpTX@mail.gmail.com>
Subject: Re: Strange behavior with templates and G++
From: Jonathan Wakely <jwakely.gcc@gmail.com>
To: gcc-help@gcc.gnu.org
Cc: Axel Freyn <axel-freyn@gmx.de>
Content-Type: text/plain; charset=windows-1252
Content-Transfer-Encoding: quoted-printable
X-IsSubscribed: yes
Mailing-List: contact gcc-help-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-help.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-help/>
List-Post: <mailto:gcc-help@gcc.gnu.org>
List-Help: <mailto:gcc-help-help@gcc.gnu.org>
Sender: gcc-help-owner@gcc.gnu.org
X-SW-Source: 2011-02/txt/msg00261.txt.bz2

On 17 February 2011 13:16, Axel Freyn wrote:
> On Thu, Feb 17, 2011 at 12:22:53PM +0000, Jonathan Wakely wrote:
>> [Moved from the gcc list]
>>
>> On 17 February 2011 11:03, Pascal Francq wrote:
>> > Hi,
>> > While compiling the following code, I got an error :
>> >
>> >
>> > template<class C> class Super
>> > {
>> > public:
>> > =A0 =A0 =A0 =A0Super(void) {}
>> > =A0 =A0 =A0 =A0void Test(C*) {}
>> > };
>> >
>> > class A
>> > {
>> > public:
>> > =A0 =A0 =A0 =A0A(void) {}
>> > };
>> >
>> > class A1 : public A
>> > {
>> > public:
>> > =A0 =A0 =A0 =A0A1(void) : A() {}
>> > };
>> >
>> > class A2 : public A
>> > {
>> > public:
>> > =A0 =A0 =A0 =A0A2(void) : A() {}
>> > };
>> >
>> > class Super2 : public Super<A1>, public Super<A2>
>> > {
>> > public:
>> > =A0 =A0 =A0 =A0Super2(void) {}
>> > };
>> >
>> > void Test(void)
>> > {
>> > =A0 =A0 =A0 =A0Super2 T;
>> > =A0 =A0 =A0 =A0A1* ptr;
>> > =A0 =A0 =A0 =A0T.Test(ptr);
>> > }
>> >
>> >
>> > The compiler gives me the following error for the Test() function:
>> > =A0 =A0 =A0 =A0error: request for member =91Test=92 is ambiguous
>> > =A0 =A0 =A0 =A0error: candidates are: =A0void Super<C>::Test(C*) [with=
 C =3D A2]
>> > =A0 =A0 =A0 =A0error: =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =
=A0void Super<C>::Test(C*) [with C =3D A1]
>> > But here the call refers clearly to the second method. The error still=
 appears
>> > if A1 and A2 do not inherit from a same root class. If I replace the c=
ode with
>> > an explicit call it works:
>> > =A0 =A0 =A0 =A0T.Super<A1>::Test(ptr)
>> > But this make the code less cleaner.
>> >
>> > Is this a problem related to a misunderstood concept from me, a wrong
>> > implementation or a technical problem of g++ ?
>>
>>
>> The ambiguity has nothing to do with templates, the code can be reduced =
to:
>>
>> class A1
>> {
>> public:
>> =A0 =A0 =A0 =A0void Test(A1*) {}
>> };
>>
>> class A2
>> {
>> public:
>> =A0 =A0 =A0 =A0void Test(A2*) {}
>> };
>>
>> class Super2 : public A1, public A2
>> {
>> };
>>
>> void Test(void)
>> {
>> =A0 =A0 =A0 =A0Super2 T;
>> =A0 =A0 =A0 =A0A1* ptr;
>> =A0 =A0 =A0 =A0T.Test(ptr);
>> }
>>
>> This is covered by 10.2 [class.member.lookup] in the C++ standard,
>> which says that a name is ambiguous if found in more than one base
>> class. =A0That makes the lookup ill-formed, before overload resolution
>> is attempted.
>
> In addition: it is not necessary to use everywhere an explicit call.
> You can import the function by adding "using"-definitions. With:
> class Super2 : public A1, public A2
> {
> =A0public:
> =A0 =A0using A1::Test;
> =A0 =A0using A2::Test;
> };
> you can call T.Test(ptr) directly, and C++ will do correct overload
> resolution -- both with and without templates

Right, in that case name lookup finds two overloads of "Test" without
looking in any base classes, then overload resolution selects the
right one.