[RESOLVED] pointer-to-member array crashes with virtual inheritance

[GeoRanger]

Hi,

I've got the following code which demonstrates a problem and am hoping someone can explain:

struct A
{
    double x[2];
    double i;
};
 
struct B : virtual public A
{
    double y[2];
};
 
int main(void)
{
    B test;
    test.x[0] = 300;
    test.i    = -5;
 
    A const * const pA = &test;
    B const * const pB = &test;
    bool      const b  = reinterpret_cast<int const *>(pA)
                      != reinterpret_cast<int const *>(pB); // As expected
 
    double (B::*pmemberBBi)    = &B::i;
    double (B::*pmemberBBx)[2] = &B::x;
    double (A::*pmemberABx)[2] = &B::x;
 
    double const * const pBBi_B = &(pB->*pmemberBBi); // Valid pointer even though pointer to derived
    double const * const pBB_B  = pB->*pmemberBBx   ; // Null pointer, but why if the previous line worked?
    double const * const pAB_B  = pB->*pmemberABx   ; // Works (pointer to base)
 
    double bbi_b = pB->*pmemberBBi    ; // Works fine even though it is pointer to derived
    double bb_b  = (pB->*pmemberBBx)[0]; // Crash, but why if the previous line worked?
    double ab_b  = (pB->*pmemberABx)[0]; // Works (pointer to base)
 
    return 0;
}

I'm wondering if this is a compiler bug. Why doesn't the pointer-to-(derived)-member work for the array if it works for the non-array?

Thanks,
GeoRanger

[Paul McKenzie]

Doesn't compile here:

http://ideone.com/TbDZg2

Regards,

Paul McKenzie

[2kaud]

Compiles and runs without any errors reported under VS2013 & VS2012. Just because it runs without a crash doesn't mean the program is correct though.

What compiler are you using?

[Paul McKenzie]

From 4.11.2 of the ANSI standard:

An rvalue of type “pointer to member of B of type cv T,” where B is a class type, can be converted to an
rvalue of type “pointer to member of D of type cv T,” where D is a derived class (clause 10) of B. If B is an
inaccessible (clause 11), ambiguous (10.2) or virtual (10.1) base class of D, a program that necessitates this
conversion is ill-formed.

Regards,

Paul McKenzie

[GeoRanger]

Thank you both for your interest and responses.

Thank you Paul McKenzie for the quote from the Standard. That's the kind of information I was hoping to find.

2kaud, I'm using Visual Studio Premium 2012. It compiles with no errors or warnings even at Warning Level 4 (except for several "local variable is initialized but not referenced" warnings). What a trap! Intuitively, it seems like it should work.

Thanks,
GeoRanger