GSoC Project Ideeas

Pau Garcia i Quiles pgquiles at
Tue Mar 29 22:02:34 CEST 2011

2011/3/26 Alexandru Tudorica <tudalex at>:
> Hi, I am student from the Polithenic University of Bucharest, Romania and
> was interested in the KDE Windows project.
> I was looking at the first proposal ("singe applications") since I saw that
> the second one was dependent on this one.
> I had a few questions about it, from what I understood you want to implement
> applications as a layer above modules (example: applications as groups of
> modules)?

I already answered in IRC but sending here in case someone else is interested:

[21:42] <pgquiles> I proposed it :-)
[21:42] <pgquiles> no, it's even easier than that
[21:43] <pgquiles> in KDE, module = group of related applications and libraries
[21:43] <pgquiles> for instance, the "kdegraphics" module is gwenview,
digikam, libkipi, ksnapshot, okular, etc
[21:44] <pgquiles> when we used subversion as our VCS, in order to
build gwenview, you had to download the source for the whole
"kdegraphics" module, then configure (= run CMake) for the module,
then you could build a single application, or the whole module
[21:45] <pgquiles> now that we have moved to git, most modules have
been split and applications are buildable on its own
[21:45] <pgquiles> for instance, if you want to build gwenview, you
only need to clone the git repository for gwenview and run cmake for
[21:45] <pgquiles> then build gwenview
[21:46] <pgquiles> so now back to KDE on Windows: most recipes for
emerge build modules (kdegraphics, kdeutils, etc) instead of
applications (gwenview, ksnapshot, etc)
[21:47] <pgquiles> so essential the GSoC proposal "Single
applications" means: "please, take the recipes for kdegraphics,
kdeutils, kdegames, etc and replace them with recipes for each one of
the applications"
[21:48] <pgquiles> it's a relatively easy GSoC. It takes quite some
time because you have to check everything with MinGW4 and MSVC2010,
but that's about it
[21:49] <tudalex> ok ,do you suggest any aplication that I can start
with? so that I can see how hard it is?
[21:49] <pgquiles> I think we can safely give up on mingw32 and
MSVC2008 for the next release
[21:49] <pgquiles> tudalex: let me check what modules have moved to git
[21:50] <tudalex> from my experience MSVC2010 complied much faster,
than mingw4 ( a few hours difference between those two on my Core 2
Duo 2Ghz CPU)
[21:51] <pgquiles> tudalex: start with kdegraphics, that one is mostly
in git now (save for okular, which is the only app still in svn, so
can treat it like a git repository: the whole module only builts one
[21:51] <pgquiles> ksnapshot still has a few problems on Windows, but
I have had no spare time to fix them :-/
[21:52] <pgquiles> tudalex: btw, the first idea and the second idea
(which was also proposed by me) are completely orthogonal, you can go
for the second one if you like it better, it does not require that the
"single apps" idea is fulfilled first
[21:53] <tudalex> is this a important ideea? I mean I've talked to
mentors on some other projects and they have told me that they are
more interested in other ideeas
[21:53] <tudalex> ah nice, I had a couple of questions about that one too
[21:53] <pgquiles> if you want to go for the "AppStore-like
installer", you can develop a fancy UI installer and for now install
the modules (even using the current code for downloads and the same
tarballs for installation)
[21:54] <pgquiles> tudalex: problem with the "single apps" idea is the
move to git is being much slower than I expected: kdeutils, kdegames,
etc have not moved yet

Pau Garcia i Quiles
(Due to my workload, I may need 10 days to answer)

More information about the Kde-windows mailing list