Re: GNUstep and GTK+

[Adrian Robert]

On Aug 6, 2004, at 2:07 PM, Quentin Mathé wrote:

> Le 6 août 04, à 06:00, Adam Fedor a écrit :
>
> > RMS has asked it is possible for GNUstep to have GTK+ bindings. Note 
> > this is not using GTK with GNUstep, but would allow someone to 
> > compile an application based on GTK with GNUstep - where GNUstep 
> > would interpret the GTK calls and instead create GNUstep objects 
> > (windows, etc).
> >
> > This would be interesting as it could potentially open up a whole set 
> > of Linux apps to run on GNUstep. THe downside is it would probably be 
> > a lot of effort. I don't have time, so could someone volunteer to 
> > look at whether this might be feasible or not?
>
> I don't see a real interest with such idea when I consider the 
> incredible effort which would need to be invested in such project and 
> then wouldn't be invested in GNUstep itself and other more important 
> related projects.

True, it would be a major effort, but certainly less effort than 
porting various applications individually.  For instance, someone 
recently mentioned wanting a version of emacs on a hypothetical gnustep 
distribution.

One needs to ask, though, what would be the benefits of having GTK 
applications in GNUstep, whether ported manually or via a GTK layer?  
If it is just having apps have the GNUstep look and feel, the same 
could be mostly accomplished more easily through the GTK themes 
mechanism.  If it is being able to launch apps via GNUstep facilities, 
we have the (somewhat maligned of late ;) app wrappers.  If it is 
services or other distributed objects-based capabilities, there might 
again be easier alternative means of accomplishing this, e.g., by 
writing some kind of GNUstep-GTK intermediary that talks DO on the one 
side and Gnome-CORBA on the other, and makes service building a simple 
matter of specifying a mapping in a property list.  (How to add a 
Services menu to a GTK app might be a trickier issue..)

In short, it seems like a little bit of intelligent effort could bring 
integration nearly as far as a brute-force GTK layer approach, and 
would leave more time for improving GNUstep to the point where people 
would rather develop with it than GTK or QT.

Adrian