I have picked up what I think could be a parsing bug on gcc 3.3.3 to do
with nested constructors being evaluated as a function definition. The
stripped down code below fails to compile. I've had a look through
Stroustrup's 'The C++ Programming Language' and googled around this but
nothing jumped out at me. Constructing using = works.
I was wondering if this is an error on my part, or is it a problem with
the parser?
Regards
Stuart
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
struct A
{
explicit A(int v) : val(v) {}
int val;
};
struct B
{
explicit B(A a) : val(a.val) {}
int val;
};
struct C
{
explicit C(B b) : val(b.val) {}
void doSomething() { }
int val;
};
int main(int argc,char** argv)
{
A a(100);
C c1=C(B(a));
c1.doSomething(); // <--- this works
C c2(B(a));
c2.doSomething(); // <--- fails here with error:
// error: request for member `doSomething' in `c2', which is of
// non-aggregate type `C ()(B)'
return 0;
}