Re: Problem with derived classes and overloaded member functions

On 3/24/10 5:02 AM, Saner Daniel wrote:
> Thank you for the quick response.
> (I hope I didn't mess up the 'reply-to')
> Unfortunately this was not the problem, I am always calling the
> function with the double dot. In fact, the error message is exactly the
> same though as if I was writing it with the dot and no 'self' argument.
> 
> I'm afraid to say that only now I noticed that the problem does in
> fact not relate to the functions being overloaded. I failed to notice
> that by chance, *all* functions of the class in questions were
> overloaded. I added a non-overloaded dummy function to the class and it
> gives the same error:

>> obj:doNothing()
> No matching overload found, candidates:
> void doNothing(PluginCLGeneric&)
> 
> I was thinking that maybe this is a problem related to how I
> register
> the object itself with Lua. The object is not created directly by
> calling the constructor from Lua, but a factory function which takes the
> name to give the object as a string argument. Calling this factory from
> Lua creates the object and then registers it as a global:
> 
> luabind::globals(L)[objectName] = createdInstance;
> 
> where 'createdInstance' is a pointer to the created object. Also,
> the
> class of the object and its functions have been registered with Luabind
> beforehand. This multi-step process was necessary due to the plugin
> architecture I am using it for. The process cannot be a complete failure
> though, as the object is accessible from Lua under 'objectName', and
> also the functions of the class have been registered. If any of these
> are not given, the error message does not reference the possible
> overload candidates anymore.
> As the exact same procedure for an object of a non-derived class
> that
> is created at startup works flawlessly, I am lead to believe that it is
> indeed a problem of inheritance, but I was still not able to find the
> culprit. Maybe Lua is confused about the class of the created object?


It shouldn't be. Everything in your post looks reasonable, so I guess
this could be a bug. Could you possibly reduce this to a failing test
that I can run?

-- 
Daniel Wallin
BoostPro Computing
http://www.boostpro.com

------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev