Should this be ambiguous or not? (implicit casts)

问题:

 struct A 
 {
     A(const A& src);
     A(const char* src);
 };
 struct B 
 {
     operator A();
     operator char*();
 };
 void test()  
 {
     B v;
     A s(v);
 }

EDG/Comeau and MSVC allows the code while GCC 4.4.4, CLANG and BCC reject it as ambiguous.

A C++ committee member replied with this (initially):

It's not ambiguous; the A(const A&) constructor is better than the A(const char*) constructor. The const A& parameter binds directly to the result of the conversion function, so the conversion sequence is considered to be a user-defined conversion followed by an identity conversion (13.3.3.1.4p1). The const char* parameter is a user-defined conversion followed by a qualification conversion, so it's worse.

Then, he followed up with this.

Actually, I was wrong. While it is true that the second conversion sequence in a user-defined conversion sequence is a tiebreaker, looking more closely at 13.3.3.2p3, the next-to-last bullet, reveals that this tiebreaker only applies if the two sequences contain the same user-defined conversion sequence, and that is not the case in this example. Because one constructor's conversion sequence uses B::operator A() and the other uses b::operator char*(), there's no tiebreaker between the two user-defined conversion sequences and they are ambiguous.

My question is this.

13.3.3.2 p3 states that,

Two implicit conversion sequences of the same form are indistinguishable conversion sequences unless one of the following rules apply.

From my understanding, the keywords are "one of the following rules". It doesn't mean that the bullet which states "same conversion sequence" overrides all the ones above. I would've thought "the rank of S1 is better than the rank of S2" would be applicable instead?

回答1:

Yes, the expected result is Ambiguity to the best of my interpretation of Clause 13.3.3.2

Matching argument 'v' of type 'B' to the parameters of either of the overloaded constructors of 'A' requires a user defined conversion. There both the sequences are of CONVERSION rank.

My interpretation is that the following quote from $13.3.3.2 applies

[...]User-defined conversion sequence U1 is a better conversion sequence than another user-defined conversion sequence U2 if they contain the same user-defined conversion function or constructor and if the second standard conversion sequence of U1 is better than the second standard conversion sequence of U2.

Both of these invoke different conversion functions in class 'B'. Therefore, I think the first condition itself is not satisfied and hence the expected result is Ambiguity as nither of the conversion sequences is better than the other.

回答2:

Disclaimer : the standard is really complex on these parts, so I may be completely wrong in my understanding.

Standard definition of the best viable function (13.3.3) :

Given these definitions, a viable function F1 is defined to be a better function than another viable function F2 if for all arguments i, ICSi(F1) is not a worse conversion sequence than ICSi(F2), and then

[...]

• the context is an initialization by user-defined conversion (see 8.5, 13.3.1.5, and 13.3.1.6) and the standard conversion sequence from the return type of F1 to the destination type (i.e., the type of the entity being initialized) is a better conversion sequence than the standard conversion sequence from the return type of F2 to the destination type.

If I understand correctly, the type of the object being constructed has its importance here, and that would make A::A(const A &) a better candidate.

---

Refer to Johannes comments to see why this answer is incorrect : this is indeed ambiguous for the reason pointed out by Chubsdad.