Re: Object Orientation in GTK

[SEGV <mlepage cgocable net>]

Havoc Pennington wrote:
> 
> On Sun, 8 Nov 1998, Matthew Ettus wrote:
> >
> > Isn't this essentially like global variables?  Just because you throw
> > them all into a structure, it is essentially the same, right?
> >
> 
> Not if you use the structure like an object. You don't declare it
> globally; you have a constructor, my_object_new(). You then come up with
> well-defined accessors and you have to have the discipline to stick the
> them, because C won't keep you from doing the wrong thing.
> 
> (Though you can use the public/private trick as seen in Gdk:
> 
>  foo.h:
> 
>   struct Foo {
>     int public_member;
>   };
> 
>  foo.c:
> 
>   struct FooPrivate {
>    struct Foo;
>    int private_member;
>   };
> 
>   void foo_method(Foo* f)
>    {
>     pf = (FooPrivate*)f;
>     ...
>    }
> 
> )
> 
> You get the idea. If you stick to the accessors and use dynamically
> allocated objects, you can have OO code, as Gtk does. There's no reason
> you have to use global variables. Though in many cases they are a fine
> thing to use.
> 
> Most of the code in Gnome CVS uses this Gtk-influenced style.
> 
> Havoc

Um, isn't this a little overboard? Why can't (the collective) we stop picking at
our scabs and just use C++?

For the last few years, whenever I use C I always end up reinventing the wheel:
objects, dynamic function calling, access protection, etc.

Why not just let the compiler do it for you? I mean, I had to laugh when I read
this email. This is so... primitive. Sure, it can be done, but why punish
ourselves?

Just thoughts from a C++ friendly developer. We now return to our regularly
scheduled programming.

-- 
SEGV    http://www.cgocable.net/~mlepage/