No more gtk? (was - Re: Gtk3 for Xfce 4.12?)

[Stephan Arts]

On Thu, May 3, 2012 at 8:30 AM, Auke Kok <auke at foo-projects.org> wrote:
> On 05/02/2012 10:54 PM, Stephan Arts wrote:
>>
>> On Thu, May 3, 2012 at 7:42 AM, Mark Trompell<mark at foresightlinux.org>
>>  wrote:
>>>>
>>>> Just picking Webkit as example, because it
>>>> is a pretty prominent one, but there are more examples to find.
>>>
>>>
>>> Before someone says (not without evidence) that xfce isn't relying on
>>> webkit. We do for example rely on libwnck, which doesn't in its latest
>>> releases support gtk2 anymore. So we're either forced to fork (xfce
>>> did that for libwnck in the past), to stay with old dependancies or to
>>> move on to gtk3. Forking is given the manpower of the project not what
>>> we should aim for.
>>
>>
>> I think you have just hit the sore-spot. One of our dependencies has
>> already stopped supporting gtk-2, this means the troubles have already
>> begun.
>>
>> I say, "Lets just port xfce to gtk-3"... the API-breakage is not on
>> the application-side, so with the exception of themes and the
>> theme-engine we can safely start the migration.
>
>
> There are some predictions one could make for Xfce's future, based on
> decisions we take now:
>
> 1) stay on gtk2. Xfce4 doesn't go anywhere, all interesting apps move to
> gtk3 or Qt, or the web. We lose developers, and users, to gnome3+.
>
> 2) go to gtk3. We're now even less of a distinctive desktop since we build
> on gtk3. The fight for reduction of footprint in gtk and gnome libraries is
> lost. Xfce4 becomes a "shell" on gnome3. Only disgruntled kernel developers
> use it (pun intended ;)). But seriously, the edge on other alternatives that
> don't build on gtk3 is lost.
>
> 3) ... break with Gtk. Choose a non-gtk toolkit and core library set that
> isn't pushed by a group of folks that we disagree with in regard. Oh, btw,
> there is such a toolkit out there. It's mature, does all sorts of nice
> database storage, animations, painting on wayland, X, directfb, etc. Even
> has a video playback framework, is focused on embedded/small systems with
> low memory, and runs fast, very fast.
>
> We'd probably lose a few folks attached to glib/gtk in the process, but,
> there's a large team of people supporting this toolkit that are lacking
> application writers, and this is where the Xfce community has a large group
> of. All seems such a nice combination of two groups with largely the same
> development goals and interests.
>
> I think most folks can figure out which toolkit/framework I'm referring to.
>
> Note: I have no particular preference, but, I really think everyone that has
> a long-term goal of supporting Xfce should consider all the alternatives,
> especially if there's a third one that is actually viable (think about how
> much work it will be to port every app to gtk3).

For the people who are wondering what toolkit Auke is referring to:
http://www.enlightenment.org/p.php?p=about/efl

And, I know this is a step not to be considered lightly, but we (I)
have complained about Gtk+ becoming more of a libgnome for years.

Gtk+ contains more and more stuff we don't need, don't want and
definitely don't use. (gtkswitch, gtkvolumeslider. gtkhsv) And with
each release, more stuff from libegg and libgnome found it's way to
Gtk+.
Now, with Gtk+3, the developers not being able to commit themselves to
a stable-API, I wonder if we should place our trust in them.

We have already seen that Gnome is moving away from our idea of a
usable desktop. And history has shown us, where gnome goes, gtk+
follows.
As things move on, more and more are we a traditional desktop
depending on a toolkit designed for a desktop going in the oposite
direction.

I am not saying that we should just start rewriting everything. But,
if there is any time at which we can (and should) look at the options
and consequences... that would be now.

Regards,

Stephan