Consider the following:

class Vector3 {
public:
	f64 x, y, z;
};

// pass an unkown userdata type to lua
template<typename T> inline void luaPushArg(lua_State * const LuaState, T arg) {
	tolua_pushusertype(LuaState, (void*)&arg, typeid(T).name() + 6);
}

void testFunction(const Vector3 &v) {
	luaPushArg(LuaState, v);
	luaCallFunction("TestFunction"); // CRASH (only sometimes though!)
	luaPop();
}

What went wrong here? Can you spot the bug? This one was a real pain in the tush. luaPushArg() would work with all my specialized types (int, float, etc) that I was passing to Lua, but when I passed classes it would sometimes crash. Turns out that luaPushArg() is taking a T arg instead of a T &arg. This means that a new copy of 'v' is created inside testFunction() when it calls luaPushArg(). luaPushArg() then pushes the newly created object onto the Lua stack. Upon luaPushArg()'s return, the pointer too the class object that was just pushed onto the lua stack is now invalidated. Sometimes it would crash, sometimes it wouldn't. This one was a real nightmare.

Hehehe . I can't wait for C11 support in GCC, that way bugs like this can be avoided using type-generic expression macros.

typeid(T).name() + 6

W.


T.


F.

I was assuming that it was to skip past name mangling output, but holy crap is that non-portable, and would cause serious problems if you have two classes with the same name.

typeid(T).name() + 6

W.


T.


F.

tolua++ needs to know the class's typename or else its treated as a void* inside lua. The "typeid(T).name + 6" skips the "class " prefix and only passes the classes unique type name ("class MyClass" -> "MyClass") which is what tolua++ expects. luaPushArg() has specialized template instances for all non-class types (thus avoiding the issue of instantiating the template with, say, an int). That way I can pass any class to lua without having to define a specialized instance of luaPushArg() for each class type. With hundreds of classes exported to lua, this saves a lot of time and code.

holy crap is that non-portable

Portability isn't an issue. If it was fixing it would be a matter of a couple of #ifdefs. Whoopdeedoo.

would cause serious problems if you have two classes with the same name.

Its not possible to have two classes with the same typename. Period. Try to compile this:

class A { };
class A { }; // error C2011: 'A' : 'class' type redefinition

not sure about the typeid resulting from that, but:

namespace {
class A {};
}
class A {};

The point was more that its not impossible to compile classes with same name.

The point was more that its not impossible to compile classes with same name.

Even the most sound code can be broken when you throw scenarios at it that are specifically designed to break it. My compiler doesn't mangle the class names nor do I use namespaces, hence neither is an issue.