Re: Dangling pointers Lua style...

Yes, that would work. Thank you.  

I was trying to think of a generic solution, such as some sort of
"ward_ptr" that threw an exception on getPointer() if it was invalid.
However the problem is - how would such a class "know" when the pointer
it contains has been free'd.  It probably could be done via, heap
debugging functions or some such, but that's just yuck. (Slow,
non-portable, etc)

A wrapper class it is methinks!  But hang on moment...
would this work??  How clever is Luabind about this...

void Destroy( ward_ptr<myclass> ptr )
{
	ptr->Destroy();
	ptr.invalidate();
}

class_<myclass, ward_ptr>("myclass")
[
	.def("Create", MakeOne ) // MakeOne would return a myclass*
	bla bla bla
	.def("Destroy", Destroy )
]

Hmmmmm.

Nigel

On Thu, 2010-09-30 at 13:17 -0400, Evan Wies wrote:
> You can try wrapping the Create/Destroy in a proxy class that uses 
> constructors/destructors instead?  It depends on how pervasive this 
> idiom is in your code.  From there, Luabind is pretty flexible with how 
> you can bind arbitrary functions to classes.
> 
> If you want to explicitly control lifetimes, check out how Corona deals 
> with a similar problem:
> http://developer.anscamobile.com/content/application-programming-guide-graphics-and-drawing#Removing_Objects_Properly
> 
> -Evan
> 
> 
> On 09/30/2010 09:28 AM, Nigel Atkinson wrote:
> > Consider a bound class without a bound constructor.  Instead elsewhere
> > there is a bound function to create and another to destroy.
> >
> > Used something like this...
> >
> > obj = CreateObject()
> >
> > -- use the obj
> >
> > DestroyObject( obj )
> >
> > The destroy might do more, but ultimately deletes the object.  Problem
> > is in Lua, we now have a dangling pointer!  obj after the destroy call
> > still points somewhere, and if you accidentally use it bad things
> > happen.  Not good if the Lua scripts are from the end user.  They can't
> > be expected to write "obj = nil" after each call. :)
> >
> > So how to automatically set obj to nil after a destroy call?
> >
> > My thoughts lean towards “All problems in computer science can be solved
> > by another level of indirection”
> >
> > In this case some sort of smart/auto pointer, and wrapper functions that
> > throw exceptions if the passed pointer points to bad data?
> >
> > What are everyone's thoughts?  I have the feeling I've missed some
> > obvious answer.
> >
> >
> > ------------------------------------------------------------------------------
> > Start uncovering the many advantages of virtual appliances
> > and start using them to simplify application deployment and
> > accelerate your shift to cloud computing.
> > http://p.sf.net/sfu/novell-sfdev2dev
> > _______________________________________________
> > luabind-user mailing list
> > [email protected]
> > https://lists.sourceforge.net/lists/listinfo/luabind-user
> 
> 
> 
> ------------------------------------------------------------------------------
> Start uncovering the many advantages of virtual appliances
> and start using them to simplify application deployment and
> accelerate your shift to cloud computing.
> http://p.sf.net/sfu/novell-sfdev2dev
> _______________________________________________
> luabind-user mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/luabind-user



------------------------------------------------------------------------------
Start uncovering the many advantages of virtual appliances
and start using them to simplify application deployment and
accelerate your shift to cloud computing.
http://p.sf.net/sfu/novell-sfdev2dev
_______________________________________________
luabind-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/luabind-user