inner workings
The book and some documentation I read say the vtbl plays a major role and the explanation they provided makes it seem very logical. Also, I can remember something like “for a down-cast or cross-cast to work, the concerned Type must be polymorphic”. But I tried an experiment where I did not involve virtual functions at all. Still, up-casting works. Can someone please explain why and how it still works in the absence of Vtables?
Upcasting works basically because in C++ any derived class contains data for its base classes. You can think of it as of onion layers: class A { int a; }; class B : public A { int b; }; class C : public B { int c; }; If you create instance of C it will likely be 3 ints in memory: a, b, c in this order, because: [[[A data] B data] C data]. If you upcast pointer to C to pointer to B or to A, it will be safe changing type of pointer, since memory of parent A (or B) inside memory of C is actually in the same address. It becomes a little bit more complex if multiple inheritance is involved, because then you actually need to shift address sometimes, but the shift is still known at compile time and it still safe. Not sure if anything of this is written somewhere in the standard, or it is implementation details.
Обсуждают сегодня