Re: About GTK+ 3.0 and deprecated things

["Mikkel Kamstrup Erlandsen" <mikkel kamstrup gmail com>]

2008/7/17 Travis Watkins <amaranth ubuntu com>:
> On Thu, Jul 17, 2008 at 6:51 AM, Kalle Vahlman <kalle vahlman gmail com> wrote:
>> I wish people would stop fighting against the "code cleanup release",
>> unless they really want to switch to an alternative widget library.
>> GTK+ isn't seeing the love it needs, and the strict API/ABI policy of
>> GTK+ 2.x mixed with "leaky" public interfaces is seen (by the people
>> doing the work) as one of the biggest problems why that is.
>>
>> If you prevent fixing that with all this stop-energy, it's not going
>> to take long for new applications start choosing something less dead
>> as their widget set of choice.
>>
>> So please let the code be revived _now_ when there's a chance it's not
>> going to be just undead after the operation and direct the energy to
>> implementing the canvas, css theming and animations and whatever. The
>> good news is that you don't need to wait for 3.x, you can start the
>> work already! The new API and its policy is already known and in SVN
>> if I'm not mistaken, so go and hack away!
>>
>
> The problem is this "code cleanup release" breaks everyone's
> applications for no reason. No one has shown a case where the current
> situation makes a fix impossible and no one has proven any interesting
> new features (other than whole new widgets, which we can do now) can
> be added to 'gtk+ 3.0' without breaking API/ABI. We're going to get
> terrible backlash from a 3.0 that has zero new features and I suspect
> any of the promised shiny bits will end up having to wait until 4.0
> anyway.
>
> In the end all we get is the ability to drop some old widgets no one
> spends time on anyway.

I have two comments:

1) How does *anything* break? I am willing to bet 10 cases of beer
that distros are going to ship gtk 2.0 and 3.0 side by side for a long
time (unless it is absolutely trivial to port all major apps). Also
with the widespread deployment of Gtk2 it is also likely to assure
that Gtk2 will bitrot in the short term.

2) I also think that the psychological aspect of this whole deal gets
way to little attention. Hackers that are not seasoned Gtk-veterans
(like yours truly) are more and more regarding Gtk as an outdated
toolkit. And you can rest assure that this is what the lay hacker
think. I walk and talk with people that think this every single day.
At work, on mailing lists, on IRC. Everywhere. It is just not cool to
use Gtk.

As I see it Gtk is in desperate need of of a cool-juice injection if
it want to continue to appeal to aspiring hackers. Talking about
myself I can attest that I am looking more and more for alternatives.
If it was not because my mind was incompatible with C++ (although I
really want to like the language) and that my attention span simply
can not cope with KDE I would probably have been a Qt consumer today.
Call me an eye-candy-junkie or what ever derogatory term you want to
apply, but people out there in the trenches are *already* seeking out
alternative toolkits. You don't have to look very far to see that, but
I shall omit any finger pointing.

As mentioned this is very much a psychological phenomenon, not
necessarily rooted in rigorous technical arguments. We need something
flashy to turn this sentiment, technical rants and ABI stability will
not cut it.

Conclusion:
Gtk does not have to break and go for 3.0. It can probably survive a
good many years yet and I will continue to use it myself. However; if
the idea is to keep aspiring hackers interested something has to be
done. This need not be a goal of Gtk and I can accept that. There are
plenty of reasons not to break, like don't upsetting key stake
holders. In the end it comes down to which goal is deemed most
important: Keep ISVs and spare-time-investors happy or motivate a new
generation of hackers to pick up, the coolness that could be, Gtk.

Cheers,
Mikkel