Re: Gnome 3.0 and implementation

[Miroslav Silovic <miro puremagic com>]

bcaccinolo idealx com wrote:

> For a good implementation of this concept could it be possible to use a
> real object language?
>  
This topic has been rehashed in the past. Surprisingly enough, it just 
keeps coming up again.

> I really don't want ot troll but I think this could be really
> interesting to use such a langage. Like Seth has said in his article on
> gnome journal,  there is activity in pygtk and gtk# communities and IMHO
> it is due to the use of a real object language.
>
>  
If gtk were written in a 'real object language' (pick any one you like), 
there would be no pygtk or gtk# - you'd only have one of them. The main 
disadvantage of 'real object languages' is their stubborn refusal to 
properly talk to each other. .NET is no improvement to this - it's still 
just one VM, and there are others. VMs also refuse to properly talk to 
each other (marshalling calls and shunting to IPC does not qualify as 
'properly', and neither does C++ style of wrapping everything into 
extern "C" stubs).

IMHO (and I'm no bigname GNOME developer, so a large grain of salt is 
advised :) ), C is just right for the low-level of GNOME. That said, GOB 
makes writing gobjects really, *really* nice. More extensive use of GOB 
could be something to look at?

   Miro