Why is this not ambiguous?

R

REH

Can some tell me why the chooses the constructor in class B over operator B
in class A? Is this not ambiguous?

Thanks.


#include <iostream>

using namespace std;

struct A;

struct B {
B() {}
B(const A&) {} // this is choosen
};

struct A {
A() {}
operator B() const {return B();} // this is not
};

int main()
{
A a;
B b;

b = a; // ambiguous?
}
 
V

Victor Bazarov

REH said:
Can some tell me why the chooses the constructor in class B over operator B
in class A? Is this not ambiguous?

I am not sure I understand the first question. The program is ill-formed
because the "b = a" _is_ in fact ambiguous.
Thanks.


#include <iostream>

using namespace std;

struct A;

struct B {
B() {}
B(const A&) {} // this is choosen
};

struct A {
A() {}
operator B() const {return B();} // this is not
};

int main()
{
A a;
B b;

b = a; // ambiguous?
}


V
 
H

Howard

REH said:
Can some tell me why the chooses the constructor in class B over operator
B
in class A? Is this not ambiguous?

Thanks.


#include <iostream>

using namespace std;

struct A;

struct B {
B() {}
B(const A&) {} // this is choosen
};

struct A {
A() {}
operator B() const {return B();} // this is not
};

int main()
{
A a;
B b;

b = a; // ambiguous?
}

I think the answer is given in Stroustrup's "The C++ Programming Language",
p277: "User-defined conversions are considered only if they are neccessary
to resolve a call." I don't have the Standard to back up that statement,
but if he's correct, then the fact that a copy-constructor exists that can
handle the job negates the need to bother looking at other alternatives.
Thus, no ambiguity is encountered.

-Howard
 
A

Alf P. Steinbach

* REH:
Can some tell me why the chooses the constructor in class B over operator B
in class A? Is this not ambiguous?

Thanks.


#include <iostream>

using namespace std;

struct A;

struct B {
B() {}
B(const A&) {} // this is choosen
};

struct A {
A() {}
operator B() const {return B();} // this is not
};

int main()
{
A a;
B b;

b = a; // ambiguous?
}

Comeau online compilation:

Comeau C/C++ 4.3.3 (Aug 6 2003 15:13:37) for ONLINE_EVALUATION_BETA1
Copyright 1988-2003 Comeau Computing. All rights reserved.
MODE:strict errors C++

"ComeauTest.c", line 18: error: more than one user-defined conversion from "A"
to
"const B" applies:
function "A::eek:perator B() const"
function "B::B(const A &)"
b = a; // ambiguous?
^

1 error detected in the compilation of "ComeauTest.c".

In strict mode, with -tused, Compile failed
 
R

REH

Alf P. Steinbach said:
* REH:

Comeau online compilation:

Comeau C/C++ 4.3.3 (Aug 6 2003 15:13:37) for ONLINE_EVALUATION_BETA1
Copyright 1988-2003 Comeau Computing. All rights reserved.
MODE:strict errors C++

"ComeauTest.c", line 18: error: more than one user-defined conversion from "A"
to
"const B" applies:
function "A::eek:perator B() const"
function "B::B(const A &)"
b = a; // ambiguous?
^

1 error detected in the compilation of "ComeauTest.c".

In strict mode, with -tused, Compile failed
That's interesting. Even compiling with all warnings, GCC failed to
complain. I would file a bug report, but I'm not sure which is correct. I
just trying it with VC 2003 ToolKit, and says:

error C2679: binary '=' : no operator found which takes a right-hand operand
of type 'A' (or there is no acceptable conversion)

Which I think is wrong, or at least misleading. I guess I will have to look
in the standard, which is hard for me to comprehend.
 
R

REH

Victor Bazarov said:
I am not sure I understand the first question. The program is ill-formed
because the "b = a" _is_ in fact ambiguous.
Sorry, that first question should have read, "Can some tell me why then
compiler chooses the constructor in class B over operator B in class A?"
Meaning, instead of complaining about the ambiguity, my compiler, given the
choice of:

1) B::B(const A&);
2) A::eek:perator B() const;

It chose option #1.
 
I

Ioannis Vranos

REH said:
That's interesting. Even compiling with all warnings, GCC failed to
complain.


C:\c>g++ -std=c++98 -pedantic-errors -Wall temp.cpp -o temp.exe
temp.cpp: In function `int main()':
temp.cpp:22: error: conversion from `A' to `const B' is ambiguous
temp.cpp:14: note: candidates are: A::eek:perator B() const
temp.cpp:9: note: B::B(const A&)

C:\c>
 
I

Ioannis Vranos

REH said:
Thanks. I assumed that "-ansi" would catch it, but it does not.

The complete g++ options that I am using may help:


-std=c++98 -pedantic-errors -Wall -O3 -ffloat-store -mtune=pentium3
 
R

REH

Ioannis Vranos said:
The complete g++ options that I am using may help:


-std=c++98 -pedantic-errors -Wall -O3 -ffloat-store -mtune=pentium3

Thanks. As another poster informed me, it was "-pedantic" that does the
trick. Though, I don't understand why you have to utilize an option to get
GCC to report this as an error.

REH
 
I

Ioannis Vranos

REH said:
Thanks. As another poster informed me, it was "-pedantic" that does the
trick. Though, I don't understand why you have to utilize an option to get
GCC to report this as an error.


GCC considers itself smarter than this, I guess. :) Or in other words, it considers that
the copy (=conversion) constructor of the type on the left that is assigned the value of
the type on the right, makes sense to be used (or in other words, it considers that type
conversions (copy constructors) have more priority than compatibility operators).

Theoretically speaking, both should have the same end result. But I think no one would
define both of them on purpose, so here we are. :)
 
M

Marcelo Pinto

Howard said:
I think the answer is given in Stroustrup's "The C++ Programming Language",
p277: "User-defined conversions are considered only if they are neccessary
to resolve a call." I don't have the Standard to back up that statement,
but if he's correct, then the fact that a copy-constructor exists that can
handle the job negates the need to bother looking at other alternatives.
Thus, no ambiguity is encountered.

-Howard

The Standard says, in 12.3-2:
User-defined conversions are applied only where they are unambiguous
(10.2, 12.3.2).

Thus, I believe it's an error to provide the two conversions, because
they are always ambigous and thus could not be applied.

On the other hand, The Standard says, in 12.3.2-1:
"... Classes, enumerations, and typedef-names shall not be declared in
the typespecifier-seq. ..."

So I believe that the conversion operator to B is invalid.

I may be interpreting The Standard wrongly, though.

Regards,

Marcelo Pinto.
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Members online

No members online now.

Forum statistics

Threads
473,744
Messages
2,569,484
Members
44,903
Latest member
orderPeak8CBDGummies

Latest Threads

Top