Re: Freshmeat stats

[Claus Schwarm <c schwarm gmx net>]

On Wed, 13 Oct 2004 12:39:59 -0500
Federico Mena Quintero <federico ximian com> wrote:

> One one of Luis's notes, there's the task of figuring out how many
> Gnome vs. KDE applications there are in Freshmeat.  I looked in their
> "Browse" pages for X11 apps, and this is what came up:
> 
> Environment: X11 Applications
> 
> Gnome: 902 projects
> GTK: 1037 projects
> KDE: 707 projects
> Qt: 452 projects
> 
> I don't know if that means anything :)  Still,
> 
> 1. Why are there more GTK projects that Gnome projects, while there
> are more KDE ones than Qt ones?  Are the Gnome libraries hard to use /
> undocumented / or what, compared to the KDE ones?
> 
> 2. We have more projects anyway.  However, the ratios between
> Gnome/Gtk and KDE/Qt are very different.
> 
> This is http://freshmeat.net/browse/229/
> 
>   Federico
> 

I have  1311 entries in the gnome-apps.berlios.de database right now,
just in case it might be interesting. This is GTK+ only as well as GNOME
apps, actively maintained as well as dead, or nearly dead. Some core
GNOME projects are missing because there's no webpage available for
them, or I simply forgot or overlooked them. Applets or themes are no
included.

Overall, a guess of roughly 1.500 projects surrounding GNOME or GTK+
seems reasonable.

Another statistic is available for Debian distributions, please see 

http://popcon.debian.org/ 

for actual data.

Installed base (from  6866 respondents):
library	installed	Used	Not used	Upgraded	No info	In Sid?
libgtk1.2	5104 		2520	1102		979		503		Yes
libgtk2.0-0	4394		2938	345		1102		9		Yes

I can upload a sorted table comparing toolkit libaries if someone's
interested.

Another statistic can be done by "apt-cache rdepends <package>". Please
see

http://gnomesupport.org/forums/viewtopic.php?t=7930

for some numbers, and how this was collected. Doing the same for the
"core" GNOME apps, will probably deliver a somewhat better picture of
the success of GTK/GNOME vs. QT/KDE. Oh, and if you're running Debian,
UserLinux or Ubuntu, please run the script and send my the results
including a note which Debian distribution it was taken from, please. 
:-)

To the questions, in my humble opinion:

About 1.) 
a.) One reason is are the already mentioned "GNOME myths". It's
mentioned in nearly every discussion on OSNews for month, probably also
Slashdot, and lots of other places but there's no benchmark available
nor any proper statement from a core hacker. Thus, the myth building
process goes on.

b.) Second reason is a capsulation of core GNOME hackers and the
projects. Blogs seldomly allow comments, the GNOME web pages are static,
bugzilla is hard to understand, and it doesn't differentiate between
feature requests and real bugs, the GNOME forum seems to be ignored by
all known hackers, but Murray Cumming (1 post), and mailing lists are
not obvious to users (a HIG term I believe).

c.) Third reason is (a sort of) different behaviour towards so-called
"third-party" or "fifth toe" apps. If these are already referred as to
"does not belong to GNOME", you shouldn't wonder why lots of people
ignore GNOME (libraries). Additionally, GNOME hackers started or
promoted projects that adress the same issues as so-called third party
app which makes those maintainers wonder why they should care about
GNOME (Tomboy being the latest example). There is no effort (that I know
of) to get those "GTK+ only" apps on board.


About 2.)
The reason, GTK+ leads as a toolkit compared to QT is the possibility to
ignore GNOME libaries. The better TK/DE ratio is due to their better
communication, and also to their different architecture, IMHO.

Another reason for the TK/DE ratio is the classification on freshmeat:
Abiword, for example, is not classified as GTK or GNOME, others are
classified in both. I don't think that QT/KDE has a similar problem. The
number of themes may additionally screw the statistic.

As usual, sorry for my bloody bad English. Please don't consider this to
be a rant or complain but a description from an outsider point of view.
:-)


Claus