Re: Empowering platform developers [Was: GUnique]

[Richard Hughes <hughsient gmail com>]

On Fri, 2006-09-22 at 20:03 +1000, Jeff Waugh wrote:
> <quote who="Richard Hughes">
> 
> > I also think one of the reasons it was not written with gtk+ as a target
> > was the "level choice" i.e. does this stuff belong in gtk+, libgnome,
> > <insert_project_ridley_module_here> or some other module.
> 
> It's really important we put a lid in this kind of confusion quickly, so we
> can breathe life back into coherent platform design and development. It is a
> serious problem when some of our best hackers don't feel empowered to design
> for a particular target: Seriously, why *wouldn't* GTK+ be the correct place
> to solve this *obviously* important use case (how many apps on our desktop
> or embedded environments do or should need this functionality - heaps)?
> 
> I'm not giving you stick here, Richard... although reading back I realise it
> may sound that way. Mostly, I'm rocking the boat to find out how we can do
> platform design and development in a more coherent way. Is the GTK+ team too
> scary? Are GNOME platform requirements incompatible with GTK+ at all? Can we
> build a better bridge between the projects? How can we break the cycle of
> bollocks solutions like libegg and so on? How do we empower platform hackers
> to make decisions and Get Things Done?

No stick taken :-) For me, is the dependency issue. Can gtk+ depend on
DBUS? If the answer is yes, then the decision is a no-brainer - put
libguniqueapp into gtk.

If the answer is no[1], then we need something higher in the stack like
libgnome, although I agree that it is not the preferred solution.

Also, I want stuff to work *now*, not in 18 months time with a gtk+
module that (for me at least) is far on the horizon. The 'temporary'
shared module gives us that.

Thanks.

Richard

[1] or maybe, or "we probably need to ifdef stuff just in case..."