Re: vfunc question

["Murray Cumming" <murrayc murrayc com>]

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi!
>
> After being able to load my C++ plugin in Anjuta I have a problem
> calling it's activate function.
>
> bool activate() is a virtual function wrapped using _WRAP_VFUNC().
> However, the C code calls klass->activate which ends up calling
> Anjuta::Plugin::activate_vfunc_callback which then calls
> Plugin::activate_vfunc().
>
> Anyway, activate_vfunc() contains
>
> bool Anjuta::Plugin::activate_vfunc()
> {
>   BaseClassType *const base = static_cast<BaseClassType*>(
>       g_type_class_peek_parent(G_OBJECT_GET_CLASS(gobject_)) // Get the
> parent class of the object class (The original underlying C class).
>   );
>
>   if(base && base->activate)
>     return (*base->activate)(gobj());
>
>   typedef bool RType;
>   return RType();
> }
>
> But this code will never ever calls the overridden activate() function
> of my derived SamplePlugin class.

An activate() function in your derived class has nothing to do with this.
To  override the "activate" vfunc (technically a function pointer in the C
struct) in a gtkmm class, you must override activate_vfunc().

The _vfunc() suffix was added in gtkmm to avoid clashes with similarly
named methods. For instance, a GTK+ widget can have a vfunc called
something and a method called something. It's not a clash in C:

struct Example
{
...
}

struct Example_Class
{
...
  void (*something)(Example*, int)
}

example_something(Example* example, double a, double b)


> Is this intended? I feel that I
> completely misunderstand some things with virtual functions because what
> I want is that klass->activate() ends up calling SamplePlugin::activate().

So your knowledge of the C API is the problem. Just override the C++
virtual method that exists.

Murray Cumming
murrayc murrayc com
www.murrayc.com
www.openismus.com