How does this work?

[LiquidE_SA]

Can someone please tell me how the following is possible? How is it possible to cast a LeafTrieNode pointer to a "NonLeafTrieNode" pointer, like this? p->ptrs[pos] = (NonLeafTrieNode*) new LeafTrieNode(suffix); The program works like this, but I don't know how it works, because "LeafTrieNode" isn't derived from "NonLeafTrieNode" "lieaftrienode.h" #ifndef LEAF_TRIE_NODE_H #define LEAF_TRIE_NODE_H class LeafTrieNode { public: LeafTrieNode(){}; LeafTrieNode(char*); private: bool leaf; char *word; friend class Trie; }; #endif "nonlieaftrienode.h" #ifndef NON_LEAF_TRIE_NODE_H #define NON_LEAF_TRIE_NODE_H class NonLeafTrieNode { public: NonLeafTrieNode(){}; NonLeafTrieNode(char); private: friend class Trie; NonLeafTrieNode **ptrs; char *letters; bool isLeaf; bool endOfWord; }; #endif -- modified at 9:54 Monday 19th September, 2005

[RichardS]

Hi, Simple answer is the same way you can cast anything else with a pointer. A pointer just points to memory, hence it can be made to "think" that it is pointing to a different object type. The new creates a pointer of type LeafTrieNode, and the cast works because you are allowed to change a pointers type. regards, Rich "Programming today is a race between software engineers striving to build bigger and better idiot-proff programs, and the Universe trying to produce bigger and better idiots. So far the Universe is winning." -- Rich Cook

[VaporTrace]

:confused: I can't see that it would with what you have displayed because there's alot missing to a completed and running example. What exactly is p-> ? If p is a member of a child class of both or a trianglular reference to a void parent member not listed, I could see where your compiler may confused. Either way, it looks like you are opening up to a memory leak.

[VaporTrace]

Hi Rich, I'm just pondering how his compiler is allowing the cross reference from one object pointer to the other unless p-> is of void that, yes just points to raw memory space by which he will need to not only delete the pointer but reclaim the memory space as well or get a memory leak. :doh:

[RichardS]

Hi VaporTrace, It is just a cast. Any pointer can be cast into any other pointer. The fact that they are unrelated is the programmers fault. Yes you would get a memory leak if you deleted the pointer as is, but the programmer caused the problem not the compiler. Take the situation:

class A {};
class B {};
A *castme = (A *) new B();

The compiler just makes castme point to the memory irrespective of the pointers original type. Remember C/C++ is not type safe so you are allowed to do things like this. Ask yourself how void pointers are created. They need to be a pointer of some type before they become a void pointer; you would have had to cast it to a void pointer first. Hope this explains it better. regards, Rich "Programming today is a race between software engineers striving to build bigger and better idiot-proff programs, and the Universe trying to produce bigger and better idiots. So far the Universe is winning." -- Rich Cook

[VaporTrace]

Duh yup, When you simplify, it's painfully clear as A* = A* to new object as: NonLeafTrieNode **ptrs = (NonLeafTrieNode*) new object; I think you know what I'm trying to say rather that ptrs[x]. That this pointer -> just kept saying "I'm in a mystery object" which doesn't matter. Err. Hate it when that happens. :doh: -- modified at 1:50 Tuesday 20th September, 2005 :suss: Ya know this dumb thing kept buggin me so I went to my compiler and guess what? It worked! But it still kept bugging me since this is not a normal thing I would try so I thought outside of the obvious why havn't I tried this just for the stupidity of it. Then I thought. :cool: class A{}; class B{ public: int j; }; int main (int argc, char* argv[]){ A* newptr = (A*) new B; newptr->j=5; return 0; } ;P Ya, I was a bit suprised when it let me do the initial A* newptr = (A*) new B; because it just doesn't feel right. Guess I'm too methodical.

[RichardS]

But you will probably find that you corrupted memory by doing the newptr->j=5; I agree this is not normal programming practice and no-one really does it unless there is a very good reason. "Programming today is a race between software engineers striving to build bigger and better idiot-proff programs, and the Universe trying to produce bigger and better idiots. So far the Universe is winning." -- Rich Cook

[Johann Gerell]

A general observation: Don't use C-style casts for non-PODs in C++. They make it far too easy to cast a mountain to hollandaise sauce. -- The Blog: Bits and Pieces -- modified at 1:55 Wednesday 21st September, 2005

[RichardS]

By the by, Johann, what is a POD?? "Programming today is a race between software engineers striving to build bigger and better idiot-proof programs, and the Universe trying to produce bigger and better idiots. So far the Universe is winning." -- Rich Cook

[Johann Gerell]

Oh, sorry. POD = Plain Ol' Data, ints, doubles and such goodies. -- The Blog: Bits and Pieces -- modified at 17:18 Saturday 24th September, 2005